WEBVTT Kind: captions Language: en 00:00:00.080 --> 00:00:03.200 And now let's move to the opening keynote, 00:00:03.640 --> 00:00:08.600 for which we are delighted to welcome Jutta Treviranus 00:00:08.680 --> 00:00:12.200 Jutta is the director of the Inclusive Design Research Center 00:00:12.200 --> 00:00:17.880 and a professor in the Faculty of Design at the OCAD University in Toronto. 00:00:18.280 --> 00:00:20.640 So, Jutta, the floor is yours. 00:00:21.760 --> 00:00:22.880 Thank you. 00:00:22.880 --> 00:00:26.960 And if you stop sharing, then I can share my slides. 00:00:27.920 --> 00:00:29.360 Thank you, Carlos. 00:00:29.360 --> 00:00:35.720 And it's a great pleasure to be able to talk to you about this important topic. 00:00:35.720 --> 00:00:38.120 I am going to 00:00:38.560 --> 00:00:41.000 just start my slides 00:00:43.080 --> 00:00:47.280 and I'm hoping that what you see is just the 00:00:47.440 --> 00:00:50.960 the primary slide, correct? 00:00:50.960 --> 00:00:51.880 Correct. 00:00:51.880 --> 00:00:54.680 Oh, wonderful. Okay. 00:00:54.680 --> 00:00:56.120 And thank you, everyone. 00:00:56.120 --> 00:00:59.440 I will voice my slides and the information in the images. 00:00:59.960 --> 00:01:03.960 And I've titled my talk first, Do no harm. 00:01:04.760 --> 00:01:07.880 I'm usually a really optimistic person, 00:01:07.880 --> 00:01:11.680 and I'm hoping to provide an optimistic message. 00:01:11.680 --> 00:01:16.520 But to realize the benefits of AI I believe we need to first recognize 00:01:16.880 --> 00:01:18.920 and take into account the harms. 00:01:19.520 --> 00:01:21.080 I'm going to limit my discussion 00:01:21.080 --> 00:01:24.120 to the harms that are specific to people with disabilities. 00:01:25.040 --> 00:01:30.040 There's a great deal of work detailing the ethical concerns of currently deployed 00:01:30.280 --> 00:01:35.120 AI from lack of representation to human bigotry, 00:01:35.480 --> 00:01:39.560 finding its way into algorithms to manipulative practices, 00:01:39.840 --> 00:01:44.640 unfair value extraction and exploitation and disinformation. 00:01:45.000 --> 00:01:47.920 I'll focus on accessibility and disability, 00:01:48.320 --> 00:01:50.600 including the recognition that disability 00:01:51.320 --> 00:01:54.760 is at the margins of all other justice deserving groups 00:01:54.760 --> 00:01:58.880 and therefore most vulnerable to the general and emerging harms, 00:01:58.880 --> 00:02:01.880 but also the potential opportunities of AI. 00:02:02.960 --> 00:02:06.200 Carlos shared a number of questions that were submitted 00:02:06.200 --> 00:02:10.600 by those of you attending today, and they are great questions. 00:02:10.760 --> 00:02:13.880 Shari and I have agreed that these will be better 00:02:13.880 --> 00:02:16.720 covered through a conversation than a presentation. 00:02:17.160 --> 00:02:21.040 So at the end of my talk, 00:02:21.720 --> 00:02:24.680 I'm going to invite Shari to talk 00:02:25.200 --> 00:02:29.720 about those particular questions, and we'll do the same at the book 00:02:29.720 --> 00:02:34.280 ending talk that Shari is giving tomorrow. 00:02:37.400 --> 00:02:41.080 So our society is plagued 00:02:41.080 --> 00:02:43.720 by more and more difficulties 00:02:45.240 --> 00:02:48.000 as the world becomes more and more complex 00:02:48.000 --> 00:02:51.320 and entangled, the choices increase in ambiguity, 00:02:51.320 --> 00:02:55.520 the risks associated with each decision become more consequential. 00:02:55.960 --> 00:02:58.760 The factors to consider in each decision 00:02:58.760 --> 00:03:04.640 more numerous, convoluted and confusing, and especially in times of crisis 00:03:04.640 --> 00:03:07.400 like we've been experiencing these last few years 00:03:07.720 --> 00:03:11.800 and in highly competitive situations where there is scarcity, 00:03:12.080 --> 00:03:15.920 AI decision tools become more and more attractive and useful. 00:03:17.640 --> 00:03:20.480 As a illustrative example, 00:03:20.480 --> 00:03:24.400 it is no wonder that over 90% of organized nations 00:03:24.760 --> 00:03:28.560 use some form of AI hiring tool, according to the U.S. 00:03:28.880 --> 00:03:31.600 Equal Employment Opportunity Commission. 00:03:32.560 --> 00:03:35.440 As work becomes less formulaic 00:03:35.440 --> 00:03:38.000 and finding the right fit becomes more difficult. 00:03:38.600 --> 00:03:42.480 They are a highly seductive tool. As an employer 00:03:42.960 --> 00:03:46.400 when choosing who to hire from a huge pool of applicants, 00:03:46.400 --> 00:03:48.520 what better way to sift through 00:03:48.520 --> 00:03:52.520 and find the gems and eliminate the potential failed choices 00:03:52.520 --> 00:03:57.320 than to use AI system with an AI tool making the decisions, 00:03:57.320 --> 00:04:01.560 we remove the risks of conflicts of interest and nepotism. 00:04:02.120 --> 00:04:06.480 What better way to determine who will be a successful candidate than to use 00:04:06.480 --> 00:04:10.520 all the evidence we've gathered from our current successful employees, 00:04:11.240 --> 00:04:15.400 especially when the jobs we're trying to fill are not formulaic. 00:04:15.680 --> 00:04:18.840 When there isn't a valid test we can devise for candidates 00:04:19.600 --> 00:04:21.560 to determine their suitability. 00:04:21.560 --> 00:04:25.920 AI can use predictive analytics to find the optimal candidates. 00:04:26.840 --> 00:04:31.360 In this way, we're applying solid, rigorous science and what would be 00:04:31.360 --> 00:04:34.400 an unscientific decision otherwise, 00:04:34.400 --> 00:04:37.520 we're not relying on fallible human intuition 00:04:39.080 --> 00:04:39.720 tools are 00:04:39.720 --> 00:04:42.880 even adding information beyond the application 00:04:42.880 --> 00:04:46.520 to rule out falsehoods or exaggerations in the applications. 00:04:46.920 --> 00:04:48.800 After all, you never know. 00:04:48.800 --> 00:04:51.280 There are so many ways to fake 00:04:51.800 --> 00:04:55.400 a work history, a cover letter, or to cheat in academia. 00:04:56.280 --> 00:05:00.640 The AI hiring tools can verify through gleaned social media data 00:05:00.640 --> 00:05:05.200 and information available on the web or through networked employment data. 00:05:05.440 --> 00:05:06.800 After all, employees 00:05:06.800 --> 00:05:10.720 have agreed to share this as part of the conditions of employment 00:05:11.040 --> 00:05:15.200 and other employers have agreed as the conditions of using the tool. 00:05:15.640 --> 00:05:18.800 If that is not enough, AI administered and processed 00:05:18.800 --> 00:05:20.880 assessments can be integrated 00:05:22.400 --> 00:05:25.040 and the tools are going beyond the practical 00:05:25.040 --> 00:05:29.120 and qualitatively determinable capacity of candidates 00:05:29.480 --> 00:05:33.600 to finding the best fit culturally to make sure that the chosen 00:05:33.600 --> 00:05:37.400 candidates don't cause friction but integrate comfortably. 00:05:37.880 --> 00:05:41.920 The tools will even analyze data from interviews to rate the socio 00:05:42.120 --> 00:05:44.480 emotional fit of candidates. 00:05:44.920 --> 00:05:48.800 If that's not satisfactory, an employer can tweak the system 00:05:49.160 --> 00:05:53.200 to add additional factors such as their favored university 00:05:53.520 --> 00:05:55.640 or to create an ideal persona. 00:05:56.000 --> 00:05:58.880 Pick an ideal employee as the model and the systems 00:05:58.880 --> 00:06:02.600 are becoming better and more sophisticated in finding a match. 00:06:03.240 --> 00:06:06.720 The same system can then guide promotion and termination, 00:06:07.080 --> 00:06:10.400 ensuring consistency of employment policies. 00:06:11.280 --> 00:06:13.840 So what's wrong with this? Science. 00:06:13.880 --> 00:06:15.920 Math. Statistical reasoning. 00:06:16.160 --> 00:06:18.840 Efficiency. Accuracy. Consistency. 00:06:19.320 --> 00:06:22.080 Better and more accurate screening for the best 00:06:22.080 --> 00:06:25.720 fit of the scientifically determined optimal employ 00:06:25.720 --> 00:06:29.480 accurate replication and scaling of a winning formula. 00:06:30.160 --> 00:06:32.600 It's a very seductive opportunity. 00:06:33.680 --> 00:06:36.320 What could be wrong 00:06:36.720 --> 00:06:38.640 for the employing organization? 00:06:38.640 --> 00:06:43.640 We arrive at a comfortable monoculture that recreates and intensifies 00:06:44.040 --> 00:06:46.200 the successful patterns of the past. 00:06:46.720 --> 00:06:50.440 With more data and more powerful analysis. 00:06:50.480 --> 00:06:53.960 The intended target becomes more and more precise. 00:06:54.560 --> 00:06:58.040 The employer finds more and more perfect fits. 00:06:58.440 --> 00:07:00.600 What is wrong with that? 00:07:00.680 --> 00:07:03.920 For the organization, what's wrong is what happens 00:07:04.080 --> 00:07:08.000 when the context changes, when the unexpected happens. 00:07:08.400 --> 00:07:11.840 A monoculture doesn't offer much adaptation, 00:07:11.880 --> 00:07:14.720 flexibility, or alternative choices. 00:07:14.920 --> 00:07:16.320 That's a visual description. 00:07:16.320 --> 00:07:19.680 I have an image showing what happened to clone potatoes 00:07:19.680 --> 00:07:25.000 during a blight that was survived by a diverse crop. 00:07:25.600 --> 00:07:28.520 Of course, we have diversity, equity and inclusion 00:07:28.520 --> 00:07:32.680 measures to compensate for discriminatory hiring and increase 00:07:32.680 --> 00:07:36.520 the number of employees from protected underrepresented groups. 00:07:37.400 --> 00:07:38.360 But even there, 00:07:39.320 --> 00:07:40.480 there will be an 00:07:40.480 --> 00:07:44.040 even greater rift between the monoculture and the candidates 00:07:44.040 --> 00:07:48.680 hired through diversity and equity programs. 00:07:48.800 --> 00:07:53.120 What happens to the candidate with a disability who would 00:07:53.120 --> 00:07:57.080 otherwise be a great fit for doing the job, 00:07:57.560 --> 00:08:00.040 when judged by these hiring systems? 00:08:01.840 --> 00:08:04.880 When AI is analyzing, sorting, filtering data 00:08:05.360 --> 00:08:08.840 about a large group of people, what does disability look like? 00:08:09.200 --> 00:08:12.280 Where is disability in a complex and tangled 00:08:12.280 --> 00:08:14.400 adaptive multivariate data set? 00:08:15.360 --> 00:08:17.560 Self-identification is often 00:08:17.560 --> 00:08:20.840 disallowed and many people don't self-identify. 00:08:21.200 --> 00:08:24.800 Even if we had a way to identify the definition 00:08:24.800 --> 00:08:27.920 and boundaries of disability are highly contested. 00:08:28.280 --> 00:08:32.120 Disability statisticians are acutely aware of some of the challenges. 00:08:33.240 --> 00:08:37.280 In any normal distribution, someone with a disability is an outlier. 00:08:37.360 --> 00:08:41.200 The only common data characteristic of disability is difference 00:08:41.200 --> 00:08:43.080 from the average or norm. 00:08:43.080 --> 00:08:45.440 People with disabilities are also more diverse 00:08:45.440 --> 00:08:47.600 from each other than people without disabilities. 00:08:48.480 --> 00:08:52.520 Data points in the middle are close together, meaning they are more alike. 00:08:52.560 --> 00:08:54.320 Data points at the periphery 00:08:54.320 --> 00:08:57.560 are further apart, meaning they are more different from each other. 00:08:57.960 --> 00:09:01.080 Data regarding people living with disabilities are spread 00:09:01.080 --> 00:09:05.520 the furthest in what I call the starburst of human needs. 00:09:07.280 --> 00:09:09.320 And as a result of this pattern, 00:09:09.320 --> 00:09:12.800 any statistically determined prediction is highly accurate 00:09:13.200 --> 00:09:17.000 for people that cluster in the middle, inaccurate, as you move from the middle 00:09:17.280 --> 00:09:20.400 and wrong, as you get to the edge of a data plot 00:09:24.160 --> 00:09:24.680 here, 00:09:24.680 --> 00:09:29.200 I'm not talking about AI's ability to recognize and translate things 00:09:29.200 --> 00:09:33.520 that are average or typical, like typical speech or text or from one 00:09:33.520 --> 00:09:38.040 typical language to another or to label typical objects in the environment 00:09:38.360 --> 00:09:42.680 or to find the path that most people are taking from one place to another. 00:09:43.200 --> 00:09:46.560 But even there, in these miraculous tools that we're using, 00:09:46.560 --> 00:09:47.960 if we have a disability, 00:09:47.960 --> 00:09:52.240 if your speech is not average or the environment you're in is not typical, 00:09:52.520 --> 00:09:55.880 AI also fails. 00:09:56.120 --> 00:09:59.760 Disability is the Achilles heel of AI 00:09:59.760 --> 00:10:03.160 applying statistical reasoning and disability. 00:10:04.160 --> 00:10:06.240 You have the culmination of 00:10:06.320 --> 00:10:11.760 diversity variability, the unexpected complexity and entanglement 00:10:12.080 --> 00:10:15.600 and the exception to every rule or determination. 00:10:17.320 --> 00:10:18.880 AI systems are used to find 00:10:18.880 --> 00:10:21.600 applicants that match predetermined optima 00:10:21.880 --> 00:10:25.560 using large data sets of successful employees and hires. 00:10:26.360 --> 00:10:30.280 The system is optimizing the successful patterns of the past. 00:10:30.320 --> 00:10:32.520 All data is from the past. 00:10:32.520 --> 00:10:35.360 The analytical power tool is honing in on 00:10:35.360 --> 00:10:38.000 and polishing the factors that worked before, 00:10:38.360 --> 00:10:41.200 and we know how much hiring there is 00:10:41.200 --> 00:10:43.280 of people with disabilities in the past. 00:10:44.480 --> 00:10:47.720 The tool is built to be biased against difference. 00:10:47.760 --> 00:10:49.760 Disability is difference. 00:10:49.760 --> 00:10:51.440 Different ways of doing the job. 00:10:51.440 --> 00:10:55.320 Different digital traces, different work and education history, 00:10:55.320 --> 00:11:00.920 different social media topics, and a tangled profile of many differences 00:11:02.360 --> 00:11:06.200 as AI gets better or more accurate 00:11:06.200 --> 00:11:10.920 in its identification of the optima, AI gets more discriminatory 00:11:10.920 --> 00:11:15.960 and better at eliminating applicants that don't match the optima in some way. 00:11:17.360 --> 00:11:20.760 The assumptions these air power tools are built 00:11:20.760 --> 00:11:24.080 upon are that scaling and replicating 00:11:24.080 --> 00:11:26.840 past success will bring about future success. 00:11:27.160 --> 00:11:31.760 Optimizing data characteristics associated with past successes 00:11:32.000 --> 00:11:36.200 increases future successes and the data characteristics 00:11:36.480 --> 00:11:40.200 that determine success need not be specified or known 00:11:40.520 --> 00:11:45.440 to the operators of the AI or the people who are subject to the decisions. 00:11:45.760 --> 00:11:49.480 And the AI cannot articulate at the moment 00:11:49.480 --> 00:11:54.280 the highly diffuse and possibly adaptive reasons behind the choices. 00:11:54.600 --> 00:11:59.280 Current AI systems cannot really explain themselves or their choices. 00:11:59.560 --> 00:12:06.920 Despite the emergence of explainable AI, how many of you have experienced tools 00:12:06.920 --> 00:12:12.800 like Microsoft Viva or many other similar tools that purport to help you 00:12:12.800 --> 00:12:16.840 with be more efficient and productive by analyzing your work habits? 00:12:17.240 --> 00:12:21.760 These surveillance systems provide more and more granular data about employment, 00:12:21.760 --> 00:12:26.360 providing intelligence about the details of the average optimal employee. 00:12:27.440 --> 00:12:29.560 The results of this AI design 00:12:29.920 --> 00:12:34.160 is that the optima will not be a person with a disability. 00:12:35.040 --> 00:12:38.680 There are not enough successfully employed persons with disability, 00:12:39.800 --> 00:12:42.040 but it is more than data gaps. 00:12:42.080 --> 00:12:46.640 Even if we have full representation of data from persons with disabilities, 00:12:47.000 --> 00:12:50.440 there will not be enough consistent data regarding success 00:12:50.440 --> 00:12:53.000 to reach probability thresholds. 00:12:53.000 --> 00:12:57.040 Even if all data gaps are filled, each pattern will still be an outlier 00:12:57.040 --> 00:13:01.400 or a minority and will lack probabilistic power in the algorithm. 00:13:03.080 --> 00:13:05.040 The same pattern is happening 00:13:05.040 --> 00:13:08.200 in all life altering difficult decisions. 00:13:09.000 --> 00:13:12.600 AI is being applied and offered to competitive academic 00:13:12.600 --> 00:13:15.840 admissions departments so you won't get admitted 00:13:16.480 --> 00:13:20.320 to beleaguered health providers in the form of medical calculators 00:13:20.320 --> 00:13:23.320 and emergency triage tools resulting 00:13:23.440 --> 00:13:26.000 in more iatrogenic death and illness. 00:13:26.320 --> 00:13:30.080 If you're different from your classification to policing 00:13:30.080 --> 00:13:34.000 to parole boards to immigration and refugee adjudications 00:13:35.360 --> 00:13:36.000 to tax 00:13:36.000 --> 00:13:40.360 auditors, meaning more taxpayers with disabilities are flagged to loans 00:13:40.360 --> 00:13:43.800 and mortgage officers, meaning people with unusual asset 00:13:43.800 --> 00:13:47.920 patterns won't get credit to security departments, meaning outliers 00:13:47.920 --> 00:13:52.000 become collateral damage. 00:13:52.320 --> 00:13:57.160 At a community level, we have evidence based investment by governments, 00:13:57.560 --> 00:14:02.400 AI guiding political platforms, public health decisions, urban planning, 00:14:02.720 --> 00:14:06.080 emergency preparedness and security programs. 00:14:06.560 --> 00:14:08.040 None will decide. 00:14:08.040 --> 00:14:11.800 With the marginalized outlier, the outliers will be marked 00:14:12.040 --> 00:14:15.080 as security risks. 00:14:15.360 --> 00:14:18.440 These are monumental life changing decisions. 00:14:18.440 --> 00:14:20.560 But even the smaller, seemingly 00:14:20.560 --> 00:14:24.600 inconsequential decisions can harm by a million cuts. 00:14:25.240 --> 00:14:27.000 What gets covered by the news? 00:14:27.000 --> 00:14:29.120 What products make it to the market? 00:14:29.480 --> 00:14:32.840 The recommended route be provided by the GPS. 00:14:33.480 --> 00:14:36.360 The priority given to supply chain processes. 00:14:36.360 --> 00:14:40.600 What design features make it to the market? 00:14:40.760 --> 00:14:45.640 Statistical reasoning that is inherently biased against difference from the average 00:14:45.920 --> 00:14:51.440 is not only used to apply the metrics, but to determine the optimal metrics. 00:14:52.800 --> 00:14:56.000 And this harm predates AI, statistical 00:14:56.000 --> 00:14:59.240 reasoning as the means of making decisions does harm. 00:14:59.480 --> 00:15:03.320 It does harm to anyone not like the statistical average 00:15:03.640 --> 00:15:06.200 or the statistically determined optima. 00:15:07.440 --> 00:15:11.480 Assuming that that we know about the what we know about 00:15:11.480 --> 00:15:14.480 the majority applies to the minority, it does harm 00:15:15.560 --> 00:15:20.320 equating truth and valid evidence with singular statistically determined 00:15:20.480 --> 00:15:23.480 findings or majority truth does harm 00:15:25.680 --> 00:15:28.680 and AI amplifies, 00:15:28.680 --> 00:15:31.160 accelerates and automates this harm 00:15:32.040 --> 00:15:34.600 and it is used to exonerate us 00:15:34.600 --> 00:15:38.800 of responsibility for this harm. 00:15:39.520 --> 00:15:42.680 We've even heard a great deal about the concern for privacy. 00:15:43.280 --> 00:15:47.600 Well, people with disability are most vulnerable to data abuse and misuse. 00:15:48.080 --> 00:15:50.400 De-identification doesn't work. 00:15:50.520 --> 00:15:53.520 If you are highly unique, you will be re-identified. 00:15:54.080 --> 00:15:57.800 Differential privacy will remove the helpful data specifics 00:15:57.800 --> 00:16:02.920 that you need to take to make the AI work for you and your unique needs. 00:16:03.160 --> 00:16:06.000 Most people with disabilities are actually forced 00:16:06.000 --> 00:16:08.920 to barter their privacy for essential services. 00:16:09.680 --> 00:16:11.720 We need to go beyond privacy. 00:16:11.720 --> 00:16:16.760 Assume there will be breaches and create systems to prevent data abuse and misuse. 00:16:17.240 --> 00:16:21.120 We need to ensure transparency regarding how data is used, by whom 00:16:21.400 --> 00:16:22.880 and for what purpose. 00:16:22.880 --> 00:16:25.360 And it's wonderful that the EU is 00:16:27.440 --> 00:16:30.560 organizing this talk because the EU is doing 00:16:30.560 --> 00:16:34.600 some wonderful measures in this regard. 00:16:36.200 --> 00:16:39.080 But wait, 00:16:39.080 --> 00:16:41.120 we're talking about a great number 00:16:41.120 --> 00:16:43.320 of harms, haven't we 00:16:44.160 --> 00:16:47.440 developed some approaches, some solutions to this? 00:16:47.960 --> 00:16:52.760 Don't we have auditing tools that detect and eliminate bias and discrimination of 00:16:53.000 --> 00:16:56.520 AI and don't we have some systems that certify 00:16:56.720 --> 00:16:59.200 whether an AI is ethical or not? 00:17:00.080 --> 00:17:03.600 Can't we test tools for unwanted bias? 00:17:03.720 --> 00:17:06.760 Unfortunately, AI auditing tools are misleading 00:17:06.760 --> 00:17:10.400 in that they don't detect bias against outliers and small minorities 00:17:10.400 --> 00:17:13.360 or anyone who doesn't fit the bounded groupings. 00:17:13.760 --> 00:17:18.200 Most AI ethics auditing systems use cluster analysis, 00:17:18.200 --> 00:17:21.920 comparing the performance regarding a bounded justice 00:17:21.920 --> 00:17:25.640 deserving group with the performance for the general population. 00:17:26.080 --> 00:17:29.600 There is no bounded cluster for disability. 00:17:30.080 --> 00:17:33.360 Disability means a diffuse and highly diverse 00:17:33.360 --> 00:17:36.400 set of differences. 00:17:36.680 --> 00:17:41.080 Those AI ethics certification systems and the industry 00:17:41.080 --> 00:17:46.000 that is growing around them raise the expectation of ethical conduct 00:17:46.840 --> 00:17:50.200 that the problem has been fixed, making it even more difficult 00:17:50.200 --> 00:17:54.400 for the individual to assert and address harm. 00:17:54.840 --> 00:17:59.120 Many of them fall prey to Cobra effects or the unintended consequences 00:17:59.120 --> 00:18:03.560 of oversimplistic solutions to complex problems or linear thinking. 00:18:04.240 --> 00:18:07.880 Falling into the rut of mono-causality when the causes are 00:18:07.880 --> 00:18:09.880 very complex and entangled, 00:18:11.760 --> 00:18:12.320 there's 00:18:12.440 --> 00:18:16.160 some helpful progress in regulatory guidance. 00:18:16.200 --> 00:18:18.880 One example is the U.S. 00:18:18.880 --> 00:18:23.000 Equal Employment Opportunity Commission, which has developed the Americans 00:18:23.000 --> 00:18:26.200 with disabilities Act and the use of software, algorithms 00:18:26.200 --> 00:18:30.280 and artificial intelligence to assess job applicants and employees. 00:18:30.320 --> 00:18:32.000 A very long title. 00:18:32.000 --> 00:18:37.000 But much of the guidance focuses on fair assessments or tests and accommodation, 00:18:37.400 --> 00:18:40.880 not on the filtering out of applicants before they are invited 00:18:40.880 --> 00:18:45.880 to take an assessment or by employers who don't use assessments. 00:18:46.440 --> 00:18:50.680 The data related suggestion is to remove the disability related data 00:18:51.160 --> 00:18:54.680 that is the basis of disability discrimination. 00:18:55.240 --> 00:18:58.520 But what we found is that the data cannot be isolated. 00:18:58.520 --> 00:19:03.080 For example, an interrupted work history will have other data effects 00:19:03.080 --> 00:19:08.000 and markers, making it hard to match the optimal pattern even when that is removed 00:19:11.000 --> 00:19:13.480 For the ethical harms there are 00:19:13.760 --> 00:19:18.320 that are common to a whole group of marginalized individuals. 00:19:18.800 --> 00:19:22.280 There are numerous AI ethics efforts emerging globally. 00:19:22.560 --> 00:19:27.320 We've tried to capture the disability relevant ones in the We Count project. 00:19:27.320 --> 00:19:32.040 These include standards, bodies, which are creating 00:19:32.040 --> 00:19:34.880 a number of standards that act as guidance. 00:19:35.240 --> 00:19:39.720 Government initiatives that are looking at the impact 00:19:39.720 --> 00:19:42.600 of their decisions using automated decision tools. 00:19:42.920 --> 00:19:46.320 Academic research units that are looking at the effects 00:19:46.320 --> 00:19:50.000 and possible approaches and think tanks and not for profits. 00:19:50.600 --> 00:19:54.280 One of the things that we found, though, is that disability is often 00:19:54.280 --> 00:20:01.760 left out of the considerations or the ethics approaches. 00:20:01.760 --> 00:20:07.040 And we as the questions that were submitted, indicate 00:20:07.480 --> 00:20:09.640 we're at an inflection point. 00:20:09.640 --> 00:20:13.280 And this current inflection point we’re at reminds me 00:20:13.280 --> 00:20:17.160 of Burke and Ornstein book, The Axemaker’s Gift, 00:20:18.360 --> 00:20:19.520 they wanted us to be 00:20:19.520 --> 00:20:22.160 aware of the Axemaker's gifts. 00:20:23.480 --> 00:20:26.600 Each time the Axemaker offered a new way 00:20:26.720 --> 00:20:29.840 to cut and control the world to make us rich 00:20:30.920 --> 00:20:34.240 or safe or invincible or more knowledgeable, 00:20:34.240 --> 00:20:38.040 we accepted the gift and used it, and we changed the world. 00:20:38.040 --> 00:20:43.960 We changed our minds for each gift, redefined the way we thought, the values 00:20:43.960 --> 00:20:47.000 by which we lived, and the truths 00:20:47.000 --> 00:20:49.640 for which we died. 00:20:54.080 --> 00:20:55.000 But to regain 00:20:55.000 --> 00:20:58.040 my optimism, even AI's potential 00:20:58.040 --> 00:21:01.080 harm may be a double edged sword. 00:21:02.200 --> 00:21:04.120 The most significant 00:21:04.120 --> 00:21:08.960 gift of a AI is that it makes manifest the harms that have been 00:21:09.520 --> 00:21:12.880 dismissed as unscientific concerns. 00:21:13.520 --> 00:21:16.560 It gives us an opportunity to step back 00:21:17.080 --> 00:21:22.240 and reconsider what we want to automate or what we want to accelerate. 00:21:22.960 --> 00:21:26.320 It makes us consider what we mean by Best 00:21:26.320 --> 00:21:30.920 Buy optimal truth, democracy, planning, efficiency, 00:21:31.280 --> 00:21:36.080 fairness, progress, and the common good. 00:21:38.680 --> 00:21:41.840 Some of the things we've done within my unit 00:21:41.840 --> 00:21:46.160 to provoke this rethinking include our inverted word cloud, 00:21:46.840 --> 00:21:49.520 which is a tiny little mechanism 00:21:49.880 --> 00:21:53.840 that conventional word cloud increases the size and centrality 00:21:53.840 --> 00:21:57.120 of the most popular or statistically frequent words. 00:21:57.440 --> 00:22:01.640 The less popular or outlying words decrease in size and disappear. 00:22:02.160 --> 00:22:04.800 We've simply inverted that behavior. 00:22:04.800 --> 00:22:08.120 The novel and the unique words go to the center 00:22:08.320 --> 00:22:11.480 and grow in size. 00:22:12.120 --> 00:22:16.520 We've been trying to provocate with models like the Lawnmower of Justice, 00:22:16.880 --> 00:22:20.720 where we take the top off the Gaussian curve or the bell curve 00:22:21.000 --> 00:22:23.960 as it might be called, to remove the privilege 00:22:24.320 --> 00:22:26.360 of being the same as the majority. 00:22:26.800 --> 00:22:33.280 So the model needs to pay greater attention to the breadth of data, 00:22:33.280 --> 00:22:37.600 and we're exploring bottom up community led data ecosystems 00:22:37.600 --> 00:22:40.680 where the members govern and share in the value of the data. 00:22:41.080 --> 00:22:44.960 This fills the gap left by things like impact investing. 00:22:45.160 --> 00:22:48.600 For example, when social entrepreneurship efforts 00:22:48.600 --> 00:22:52.280 that are supposedly addressing these problems can't scale 00:22:52.280 --> 00:22:56.400 a single impactful formula sufficiently to garner support. 00:22:56.760 --> 00:23:02.120 It also works well to grow knowledge of things like rare illnesses that 00:23:02.120 --> 00:23:06.240 won't garner a market for the treatments and therefore are not invested in 00:23:09.240 --> 00:23:09.960 And create, 00:23:09.960 --> 00:23:14.920 we're also creating tools to reduce harm by signaling when a model will be wrong 00:23:14.920 --> 00:23:18.560 or unreliable, because the evidence based guidance is wrong 00:23:18.880 --> 00:23:20.880 for the person being decided about. 00:23:21.640 --> 00:23:24.920 Here we're using a tool called the Data Set Nutrition label 00:23:25.280 --> 00:23:31.480 that gives information about what data is used to train the model. 00:23:31.480 --> 00:23:35.640 But back to the axemaker’s gifts and the opportunity to reconsider 00:23:35.640 --> 00:23:36.800 where we're going. 00:23:36.800 --> 00:23:41.880 From a complexity theory perspective, where I think we're collectively stuck 00:23:41.880 --> 00:23:46.680 on a local optima and unable to unlearn our fundamental assumptions 00:23:46.920 --> 00:23:50.120 and approach approaches to find the global optimal. 00:23:51.080 --> 00:23:53.880 And I believe there is a global optima 00:23:55.280 --> 00:23:56.720 At the moment, 00:23:56.720 --> 00:24:00.440 as a society, we believe, or we act like we believe, 00:24:00.760 --> 00:24:03.080 to succeed, we need to do what we've been doing 00:24:03.080 --> 00:24:06.800 more effectively, efficiently, accurately and consistently. 00:24:07.280 --> 00:24:12.120 We're hill climbing, optimizing the patterns of the past, eroding the slope. 00:24:12.560 --> 00:24:16.280 For anyone following us, we need to stop doing 00:24:16.280 --> 00:24:20.120 the same things more efficiently and potentially reverse course. 00:24:21.280 --> 00:24:23.720 I've been considering the many local optima 00:24:24.520 --> 00:24:27.560 we keep hill climbing, not just statistical reasoning 00:24:27.560 --> 00:24:32.800 that finds a single winning answer, not just winner takes all zero sum game 00:24:32.800 --> 00:24:38.120 capitalism and economic growth at all costs, but also majority rules. 00:24:38.120 --> 00:24:40.040 All or nothing decisions. 00:24:40.040 --> 00:24:43.520 And even in our community, this accessibility community, 00:24:43.840 --> 00:24:47.000 the notion of a single checklist of full accessibility 00:24:47.400 --> 00:24:52.000 for a group of hugely diverse people, many of whom are not represented 00:24:52.240 --> 00:24:55.360 when coming up with the list, 00:24:55.400 --> 00:24:58.800 the people closest to the bottom are more diverse, 00:24:59.120 --> 00:25:02.840 closest to the path we need to follow to find the global optima. 00:25:03.280 --> 00:25:07.720 Less invested in current conventions, we need to diversify 00:25:07.720 --> 00:25:10.080 and learn to use our complementary skills 00:25:10.320 --> 00:25:13.040 and learn from people who are currently marginalized. 00:25:13.680 --> 00:25:18.960 Even in this community, focused on accessibility. 00:25:18.960 --> 00:25:24.680 Because if anyone knows, we know that it is at the margins or outer 00:25:24.680 --> 00:25:29.200 edge of our human starburst that we find the greatest innovation 00:25:29.200 --> 00:25:33.160 and the weakest and the weak signals of crisis to come. 00:25:33.760 --> 00:25:38.360 This is where you feel the extremes of both the opportunities and the risks. 00:25:41.120 --> 00:25:45.000 One of the emerging uncertainties that holds 00:25:45.000 --> 00:25:48.760 both greater opportunities and risks is generative AI. 00:25:49.520 --> 00:25:52.240 What are the implications if you have a disability? 00:25:52.280 --> 00:25:54.320 What will it do for accessibility? 00:25:55.080 --> 00:25:59.200 I'm sure you've heard about tools like GPT, ChatGPT, 00:25:59.800 --> 00:26:03.840 stable infusion and various word versions of Dall-E 00:26:05.480 --> 00:26:08.480 Midjourney and other emerging tools. 00:26:08.520 --> 00:26:11.840 Even today there's new announcements regarding new tools. 00:26:12.280 --> 00:26:16.720 These tools do not rely on purely on statistical reasoning. 00:26:17.080 --> 00:26:20.360 They can transfer learning from context to context. 00:26:20.760 --> 00:26:24.320 They use new processes called transformers 00:26:24.320 --> 00:26:27.080 that can pivot to new applications, 00:26:27.640 --> 00:26:31.640 but they can also create convincing and toxic lies. 00:26:31.640 --> 00:26:34.520 People with disabilities tend to be most vulnerable 00:26:34.760 --> 00:26:37.800 to the misuse and abuse of toxic tools. 00:26:38.520 --> 00:26:40.880 I'm going to invite Shari to help me discuss 00:26:40.880 --> 00:26:44.800 these emerging possibilities. 00:26:50.480 --> 00:26:51.560 Hello. 00:26:51.560 --> 00:26:52.960 Hello, everybody. 00:26:52.960 --> 00:26:54.600 I'm Shari Trewin. 00:26:54.600 --> 00:26:57.120 I'm from Google. 00:26:57.440 --> 00:27:00.120 And a middle aged white woman 00:27:00.120 --> 00:27:04.120 with lots of smile lines on my face. 00:27:09.400 --> 00:27:15.280 So, Jutta, you’ve given us a lot to think about there. 00:27:16.800 --> 00:27:21.640 I wonder if we might start off where you ended there. 00:27:21.640 --> 00:27:25.000 Talking a little bit about generative A.I. 00:27:25.040 --> 00:27:28.280 models and language models 00:27:29.360 --> 00:27:30.200 and the 00:27:30.200 --> 00:27:34.760 they're trained on large corpora of data that may not reflect 00:27:35.120 --> 00:27:38.920 the the moral values that we would like 00:27:39.120 --> 00:27:41.360 our models to incorporate. 00:27:42.280 --> 00:27:45.080 So one question I think would be interesting for us to talk about 00:27:45.080 --> 00:27:50.200 is can we teach these large language models or generative A.I. 00:27:50.520 --> 00:27:52.960 to apply these moral values, 00:27:53.360 --> 00:27:56.880 even though the the very large datasets may not 00:27:58.040 --> 00:28:00.680 represent that? 00:28:00.760 --> 00:28:02.600 That's a great question. 00:28:02.600 --> 00:28:06.240 And in thinking about how that might be done. 00:28:07.480 --> 00:28:08.120 One of 00:28:08.120 --> 00:28:11.840 the dilemmas is that we may need to find a way 00:28:11.840 --> 00:28:15.720 to quantify complex, abstract, qualitative values. 00:28:16.760 --> 00:28:20.600 And in that process, will that reduce these values? 00:28:21.320 --> 00:28:24.400 I mean, deep learning lacks judgment. 00:28:24.440 --> 00:28:28.400 Humans sort of value human judgment that isn't quantitative. 00:28:29.120 --> 00:28:31.960 Perhaps one way to start is 00:28:31.960 --> 00:28:34.640 by recognizing human diversity 00:28:35.760 --> 00:28:38.720 and the diversity of contexts. 00:28:38.720 --> 00:28:41.000 There is a lot of talk about 00:28:41.960 --> 00:28:44.080 individualizing applications 00:28:44.440 --> 00:28:47.280 without making the costs exorbitant 00:28:47.760 --> 00:28:50.480 and to the people that need them. 00:28:51.200 --> 00:28:55.120 The irony, of course, in that is that the people that need 00:28:55.120 --> 00:28:59.360 that type of individualization the most are also most likely 00:28:59.360 --> 00:29:01.960 to be the people that can't afford it. 00:29:02.720 --> 00:29:06.760 And I think it's not yet known. 00:29:06.760 --> 00:29:08.960 Can we do that? 00:29:08.960 --> 00:29:11.960 And of course, there's been surprising advances 00:29:11.960 --> 00:29:17.120 in all sorts of different areas with respect to AI and generative AI. 00:29:17.960 --> 00:29:22.560 But I think this is the issue of values 00:29:22.760 --> 00:29:27.400 and shared values and the the articulation 00:29:27.640 --> 00:29:33.760 and making mechanizable, because, of course, we're talking about a machine 00:29:33.760 --> 00:29:38.280 and mechanization values that that we have difficulty 00:29:38.280 --> 00:29:42.640 even fully expressing is it's quite a challenge. 00:29:42.680 --> 00:29:43.960 What do you think, Shari? 00:29:45.320 --> 00:29:45.880 Now, I 00:29:45.880 --> 00:29:48.440 think that you've hit a really good point there with 00:29:49.040 --> 00:29:52.440 can we express or can we measure 00:29:53.120 --> 00:29:57.400 whether a model meets our values or whether 00:29:58.760 --> 00:30:01.640 we think it is free from bias 00:30:01.640 --> 00:30:04.360 or as free from bias as we can make it? 00:30:05.120 --> 00:30:06.440 Do we know? 00:30:06.440 --> 00:30:07.880 Do we know how to evaluate? 00:30:07.880 --> 00:30:10.520 That I think is an important question. 00:30:11.480 --> 00:30:15.320 And some of the steps that that often get missed 00:30:15.320 --> 00:30:19.600 when creating a system that uses AI might 00:30:21.280 --> 00:30:24.680 that might help with that would be starting off 00:30:24.680 --> 00:30:28.600 from the beginning by thinking about who are the people 00:30:29.000 --> 00:30:34.000 who might be at risk, what are the issues that might be in the data? 00:30:34.000 --> 00:30:39.800 What historical biases might the data represent and include? 00:30:40.040 --> 00:30:44.280 And then actively working with members of those communities 00:30:44.640 --> 00:30:48.960 to understand how are we going to measure fairness here? 00:30:49.080 --> 00:30:50.880 How are we going to measure bias? 00:30:50.880 --> 00:30:52.200 What's our goal? 00:30:52.200 --> 00:30:54.280 And how are we going to test? 00:30:54.280 --> 00:30:57.560 How are we going to know when we've achieved our goal? 00:30:59.160 --> 00:31:01.160 So I think there's some 00:31:01.400 --> 00:31:04.040 progress that can be made in the design process 00:31:04.040 --> 00:31:08.560 and thinking about the larger system that we're embedding AI in, 00:31:09.640 --> 00:31:13.960 everything doesn't have to be built into one AI model 00:31:14.480 --> 00:31:18.640 and we can augment models, We can build systems around models 00:31:18.800 --> 00:31:22.040 that take into account their limitations and 00:31:23.360 --> 00:31:26.160 create a better overall whole system. 00:31:27.080 --> 00:31:31.680 So thinking about what the models are currently trained on 00:31:31.680 --> 00:31:35.360 and just the masses of data that are used to build the models 00:31:36.080 --> 00:31:39.360 and the training data is rife 00:31:39.360 --> 00:31:42.400 with discrimination against difference. 00:31:42.400 --> 00:31:46.520 So how do we how do they unlearn? 00:31:46.560 --> 00:31:51.920 I mean, this is it's sort of it matches some of the training 00:31:51.920 --> 00:31:55.400 that I do within my program and that 00:31:56.720 --> 00:32:00.000 students have been socialized with very similar things. 00:32:00.000 --> 00:32:03.600 And then often the issue is not learning. 00:32:03.600 --> 00:32:09.200 The issue is unlearning, like how do you remove those those unconscious 00:32:10.520 --> 00:32:13.480 habituated values that that 00:32:14.600 --> 00:32:18.000 are so embedded in our learning systems? 00:32:18.360 --> 00:32:23.240 So it is I agree that there is is huge 00:32:23.240 --> 00:32:28.040 opportunity, especially with more context aware systems 00:32:28.480 --> 00:32:31.400 and maybe what we need to pursue 00:32:31.400 --> 00:32:34.920 is even to address things like 00:32:35.960 --> 00:32:39.760 privacy and the need to swim against this massive 00:32:39.760 --> 00:32:43.320 amount of data that is not applicable to you 00:32:43.520 --> 00:32:46.440 is on device 00:32:46.440 --> 00:32:51.080 personalized or not personalized, because personalized is a term 00:32:51.080 --> 00:32:54.440 that's also sort of been hijacked to mean 00:32:54.440 --> 00:32:57.640 cushioning but individualized. 00:32:57.640 --> 00:33:02.080 Let's use that term system that takes your data 00:33:02.080 --> 00:33:04.840 and creates a bottom up picture 00:33:05.080 --> 00:33:07.000 of what is needed. 00:33:07.880 --> 00:33:10.000 Yeah, I think, you know, there 00:33:10.200 --> 00:33:15.160 there is definitely interesting avenues to explore with transfer learning. 00:33:15.160 --> 00:33:19.080 And can we take a model that's been trained on data 00:33:19.080 --> 00:33:23.360 and has learned some of the concepts of the task that we want, but 00:33:23.600 --> 00:33:26.760 maybe we'd like it to unlearn some of the things that it's learned. 00:33:27.120 --> 00:33:30.880 Can we use techniques like transfer learning to layer on top 00:33:31.360 --> 00:33:34.640 and and teach the model and direct the model 00:33:35.920 --> 00:33:38.200 more in the direction that we want? 00:33:38.200 --> 00:33:41.720 And the I think the the hopeful thing about that 00:33:41.720 --> 00:33:44.600 is that it needs orders of magnitude less data 00:33:45.240 --> 00:33:47.800 to to train such a model. 00:33:47.800 --> 00:33:50.240 And so that makes it 00:33:51.000 --> 00:33:53.720 a little more achievable, a little less daunting for 00:33:54.240 --> 00:33:57.680 for the community to take on. 00:33:59.920 --> 00:34:00.640 Yeah. 00:34:01.880 --> 00:34:03.960 Do you think that current 00:34:04.200 --> 00:34:07.960 regulation systems are really up to the task of 00:34:09.160 --> 00:34:11.920 regulating current and emerging A.I. 00:34:12.000 --> 00:34:14.920 and preventing the kinds of harms that you've been talking about? 00:34:16.000 --> 00:34:18.480 No. A simple answer. 00:34:18.560 --> 00:34:20.480 I don't think so. 00:34:20.480 --> 00:34:23.360 I mean, there's so many issues. 00:34:23.360 --> 00:34:27.080 Laws and policies are developed at a much slower pace 00:34:28.400 --> 00:34:28.880 They’re. 00:34:28.880 --> 00:34:31.360 We're dealing with an uncertain 00:34:31.360 --> 00:34:34.480 very, very quickly moving, quickly adapting area. 00:34:35.120 --> 00:34:38.480 And when laws are 00:34:40.360 --> 00:34:42.520 well, they need to be testable. 00:34:42.520 --> 00:34:45.200 And so in order to be testable, we have to create 00:34:45.200 --> 00:34:48.920 these static rules that can be tested, which means we have to be 00:34:50.200 --> 00:34:51.560 fairly specific as 00:34:51.560 --> 00:34:54.400 opposed to sort of general and abstract. 00:34:55.000 --> 00:34:59.120 And that tends to lead us towards one size fits, 00:34:59.120 --> 00:35:03.800 one criteria which we know are are not great. 00:35:03.920 --> 00:35:07.800 If we're trying to design for diversity or encourage diversity. 00:35:08.760 --> 00:35:12.200 I think we, one of the things we need to innovate in 00:35:12.200 --> 00:35:17.560 is the regulatory instruments that we can use here. 00:35:18.400 --> 00:35:21.600 What's your thinking about this? 00:35:21.600 --> 00:35:24.000 Yeah, I think some of those 00:35:24.000 --> 00:35:27.880 regulatory instruments that we have do apply. 00:35:28.280 --> 00:35:31.240 So if you're a company that's using a 00:35:31.560 --> 00:35:34.520 a AI system in screening job 00:35:34.520 --> 00:35:37.760 applicants, 00:35:37.760 --> 00:35:40.880 the disability discrimination laws still apply to you. 00:35:41.320 --> 00:35:43.680 Somebody can still bring a lawsuit against you 00:35:44.720 --> 00:35:48.920 saying that your system discriminated against them and you are still liable 00:35:49.400 --> 00:35:53.680 to defend against that and to watch out 00:35:53.680 --> 00:35:56.680 for those kinds of issues. 00:35:56.800 --> 00:36:01.160 So in in some ways, there are important pieces 00:36:01.520 --> 00:36:06.440 that we need in place that can be used to tackle problems 00:36:06.440 --> 00:36:09.800 introduced when AI systems are introduced. 00:36:10.520 --> 00:36:11.760 But then in other ways, 00:36:12.760 --> 00:36:14.400 there was a lot more of a gray area. 00:36:14.400 --> 00:36:16.720 When the technology is not making 00:36:17.840 --> 00:36:19.760 discriminatory decisions, but 00:36:19.760 --> 00:36:22.400 it still might make harmful mistakes 00:36:22.760 --> 00:36:25.520 or that mislead people 00:36:26.000 --> 00:36:29.400 or that people are relying on it for. 00:36:29.400 --> 00:36:34.400 And so, you know, if anybody here has a legal background, I would love to hear 00:36:35.480 --> 00:36:38.720 their take as well on 00:36:39.000 --> 00:36:42.320 how well do current consumer protections apply. 00:36:42.520 --> 00:36:45.080 For example, if you're using an AI. 00:36:45.880 --> 00:36:49.240 I've become aware of and worried 00:36:49.240 --> 00:36:52.440 about the people for whom the law isn't adequate. 00:36:52.960 --> 00:36:57.680 So the fact that we have a law, the fact that we supposedly have measures 00:36:57.680 --> 00:37:02.240 that prevent abuse or unethical practice, 00:37:02.640 --> 00:37:05.320 if that if you are still 00:37:05.760 --> 00:37:10.040 being treated unethically, then it makes it even harder for you. 00:37:10.520 --> 00:37:16.160 So I think the the measures that we do have, the regulations that we do 00:37:16.160 --> 00:37:20.000 have have to have some way of continuously 00:37:20.480 --> 00:37:23.600 being iterated upon so that we can catch 00:37:24.560 --> 00:37:27.080 the individuals that are not included. 00:37:27.080 --> 00:37:29.760 And we have to recognize that 00:37:29.760 --> 00:37:33.320 our supposed solutions are actually not solutions, 00:37:33.320 --> 00:37:36.040 that this is never fixed, that it's 00:37:36.320 --> 00:37:38.640 it requires this ongoing vigilance. 00:37:39.200 --> 00:37:45.240 And so the yeah, there's that there's much more to say about that. 00:37:45.240 --> 00:37:46.160 But yes, you're right. 00:37:46.160 --> 00:37:48.680 It would be great to hear from any 00:37:49.520 --> 00:37:55.640 anyone with a legal background. 00:37:55.640 --> 00:37:59.080 Yeah. 00:37:59.080 --> 00:38:02.760 Let's maybe talk a little bit about it's 00:38:03.080 --> 00:38:05.280 a bit more about generative AI 00:38:05.840 --> 00:38:09.320 that you mentioned at the end there. 00:38:09.320 --> 00:38:11.800 Know it. It 00:38:12.000 --> 00:38:17.120 produces very plausible convincing 00:38:19.640 --> 00:38:22.400 statements when when asked the question, 00:38:22.640 --> 00:38:26.360 but it also very plausibly and convincingly, 00:38:26.360 --> 00:38:31.600 completely makes things up and isn't always reliable. 00:38:31.600 --> 00:38:34.720 And in fact right now is not connected 00:38:35.040 --> 00:38:38.080 to any form of ground truth or 00:38:39.160 --> 00:38:41.600 able to assess the accuracy of what makes 00:38:42.880 --> 00:38:45.080 the one question I think that's interesting is 00:38:45.320 --> 00:38:48.680 will this technology reach a stage where 00:38:50.280 --> 00:38:52.480 it can support 00:38:52.480 --> 00:38:58.160 the kinds of decisions that we're using, statistical reasoning for right now? 00:38:58.160 --> 00:39:02.000 Eventually, obviously, right now it's it's not there yet. 00:39:03.600 --> 00:39:03.920 Yeah. 00:39:03.920 --> 00:39:08.680 And it's interesting because just recently there have been the announcements 00:39:09.120 --> 00:39:12.240 of these systems being used for medical guidance 00:39:13.520 --> 00:39:15.440 using large 00:39:15.440 --> 00:39:19.960 language models to come up with answers to your medical questions, 00:39:20.440 --> 00:39:23.600 which of course is 00:39:23.600 --> 00:39:24.480 quite. 00:39:25.040 --> 00:39:27.360 Yeah, it'll be interesting to see what happens. 00:39:28.040 --> 00:39:31.160 It's scary, I think is exactly scary. 00:39:31.400 --> 00:39:36.560 And what about the the the medical advice given to someone where 00:39:37.320 --> 00:39:42.040 within the data set that is provided, there isn't a lot of advice. 00:39:42.040 --> 00:39:44.560 So that given that the system does 00:39:45.200 --> 00:39:49.480 I mean ask any of the LLMs or the chat bots 00:39:49.880 --> 00:39:53.400 how confident they are in their their answers. 00:39:53.400 --> 00:39:54.160 They'll answer that 00:39:54.160 --> 00:39:57.320 they are confident because there isn't a sense of 00:39:58.520 --> 00:40:00.400 what is the the risk level, 00:40:00.400 --> 00:40:05.000 what is the the confidence level of this particular response. 00:40:05.000 --> 00:40:10.280 There there's no self-awareness of what is wrong 00:40:10.280 --> 00:40:13.680 and what is right and what is the context that is in front of me. 00:40:14.560 --> 00:40:16.640 Yeah, I think that's actually a great opportunity. 00:40:16.640 --> 00:40:21.000 There to explore whether we can enable 00:40:22.520 --> 00:40:25.360 models a little better to know what they don't know, 00:40:26.000 --> 00:40:29.320 to know when the case with right now 00:40:29.840 --> 00:40:35.560 isn't well-represented in their models or maybe an outlier case. 00:40:35.880 --> 00:40:38.000 That they 00:40:39.920 --> 00:40:41.120 should perhaps pass on 00:40:41.120 --> 00:40:45.880 to some other form of decision making or at least convey 00:40:46.520 --> 00:40:48.560 the less confidence 00:40:48.960 --> 00:40:53.200 in the voice of their, you know, 00:40:53.480 --> 00:40:55.640 I think generative AI today 00:40:56.560 --> 00:41:00.440 gives us a glimpse of the future and the kinds of interactions 00:41:00.680 --> 00:41:05.120 that are possible, the kinds of ways we might interact with technology 00:41:05.440 --> 00:41:07.000 in the future. 00:41:07.000 --> 00:41:11.000 You know, clearly there's a research priority to ground it better. 00:41:11.040 --> 00:41:12.640 In truth, and 00:41:13.880 --> 00:41:14.360 it needs to 00:41:14.360 --> 00:41:17.840 be much more reliable, much more trustworthy, much more accurate. 00:41:18.320 --> 00:41:24.680 But then you go to the it can support serious applications. 00:41:24.680 --> 00:41:28.160 And the idea of using it to get medical advice just is 00:41:28.840 --> 00:41:32.000 that that's a very, very scary 00:41:32.360 --> 00:41:35.120 because it is so eloquent 00:41:35.840 --> 00:41:40.320 that it's immediately trustworthy and it gets enough things 00:41:40.320 --> 00:41:44.200 right that we begin to trust it very quickly. 00:41:44.200 --> 00:41:46.600 And so in some ways, the the 00:41:47.600 --> 00:41:50.680 the advances that have been made are 00:41:52.760 --> 00:41:54.440 it's it's so good 00:41:54.440 --> 00:41:58.720 that it really highlights the dangers, 00:41:58.720 --> 00:42:07.560 I think, more effectively. 00:42:07.560 --> 00:42:10.120 So, yeah, I think that's interesting to think about 00:42:10.120 --> 00:42:12.920 what would a human A.I. 00:42:13.000 --> 00:42:17.080 interaction look like in the future? 00:42:17.080 --> 00:42:20.640 Would would we need to change train 00:42:21.720 --> 00:42:26.600 or was one to divide by bias 00:42:26.840 --> 00:42:31.080 and kind of work with a larger language 00:42:31.080 --> 00:42:34.360 model to adapt responses? 00:42:34.360 --> 00:42:35.600 Would we you know how 00:42:37.120 --> 00:42:40.040 automatic image description has sort of evolved? 00:42:40.040 --> 00:42:41.360 At first it would 00:42:41.360 --> 00:42:45.440 you know, we would throw out words and that might be in the picture. 00:42:45.440 --> 00:42:48.440 And sometimes it was right and sometimes it would be wrong. 00:42:48.800 --> 00:42:53.240 And now you see these generated alternative text 00:42:53.720 --> 00:42:58.320 being phrased in a way that conveys the uncertainty. 00:42:58.320 --> 00:43:01.800 So could be a tree or something like that. 00:43:01.800 --> 00:43:05.120 And I think the large language models could do something 00:43:05.120 --> 00:43:08.840 similar to 00:43:08.840 --> 00:43:11.360 reduce the chances of misleading people. 00:43:11.360 --> 00:43:15.440 So they might say things like many people seem to think 00:43:16.040 --> 00:43:19.280 blah, blah, blah, or get better at citing sources. 00:43:19.520 --> 00:43:23.240 I think there's a lot of ways that we can use 00:43:23.240 --> 00:43:25.640 these in direct research to, 00:43:27.160 --> 00:43:32.480 you know, to overcome some of the really obvious failings that are there right now. 00:43:32.480 --> 00:43:38.280 But the limitations that we currently have, 00:43:38.280 --> 00:43:44.040 Mark Urban has shared in in the chat that I can see 00:43:45.240 --> 00:43:48.040 from the US government regulatory side, 00:43:48.560 --> 00:43:51.160 much of the current laws 00:43:51.160 --> 00:43:56.400 or regulations for access to government services are about the technical 00:43:56.400 --> 00:44:00.360 accessibility of the interfaces rather than the more 00:44:01.560 --> 00:44:06.080 AI focused questions around system exclusion or mismatch. 00:44:06.080 --> 00:44:07.440 So that's 00:44:08.480 --> 00:44:14.400 coming back to our point about the regulatory systems. 00:44:14.400 --> 00:44:18.760 And I just noticed that Mike Calvo says what a Debbie Downer 00:44:19.400 --> 00:44:23.120 my talk is, which 00:44:23.120 --> 00:44:25.520 I think by design, 00:44:25.520 --> 00:44:28.320 we decided between Shari and I that 00:44:28.760 --> 00:44:31.360 I would provide the warnings 00:44:31.480 --> 00:44:34.360 and then Shari would provide optimism. 00:44:36.120 --> 00:44:38.480 I say I get the best of their. 00:44:41.480 --> 00:44:44.120 I think there are 00:44:44.120 --> 00:44:47.480 quite a few questions in the question and answer panel. Yes. 00:44:47.680 --> 00:44:51.040 And so maybe what we should do is... 00:44:51.160 --> 00:44:52.640 Over there's so many things 00:44:52.640 --> 00:44:56.480 to explore with the emerging models and so many uncertainties, 00:44:56.800 --> 00:45:03.680 but there's some great questions there as well. 00:45:04.320 --> 00:45:06.000 Yeah, How about 00:45:10.520 --> 00:45:13.880 jumping around on me... 00:45:13.880 --> 00:45:17.120 So many new questions. 00:45:17.120 --> 00:45:22.640 So I know this is not in the right order, but 00:45:24.240 --> 00:45:26.560 as people are adding questions, they're kind of 00:45:26.560 --> 00:45:31.640 they're kind of jumping. 00:45:31.640 --> 00:45:33.640 Okay, so Bruce Bailey is asking. 00:45:33.640 --> 00:45:35.440 He says, Fantastic keynote. 00:45:35.440 --> 00:45:39.200 Please expound on personalization having been hijacked 00:45:39.440 --> 00:45:42.320 to mean cushioning 00:45:42.440 --> 00:45:44.840 time and perspective as new. 00:45:44.840 --> 00:45:45.360 Sure. 00:45:45.360 --> 00:45:48.560 Yeah, I can definitely talk about that. 00:45:48.560 --> 00:45:50.480 I mean, one of the ways in which 00:45:50.480 --> 00:45:54.480 we recognize that we're all diverse, and especially 00:45:55.000 --> 00:45:58.520 if you have a disability, you're diverse from other people with disabilities 00:45:58.920 --> 00:46:00.360 and that our needs are there for 00:46:00.360 --> 00:46:04.160 a diverse has been to look at how do we personalize. 00:46:04.160 --> 00:46:08.600 But personalization has been used 00:46:08.720 --> 00:46:11.360 as a term to look at 00:46:13.160 --> 00:46:15.560 using recommender engines, using 00:46:15.560 --> 00:46:18.160 various ways in which we’re offered 00:46:18.600 --> 00:46:23.840 only information and recommendations from people like us, 00:46:24.200 --> 00:46:28.800 which of course removes any dissonance and any diverse 00:46:28.800 --> 00:46:33.480 thinking and our exposure to alternative views and perspectives. 00:46:34.160 --> 00:46:38.280 And to some extent it causes us to 00:46:39.040 --> 00:46:43.120 it causes greater polarization because we're also 00:46:44.000 --> 00:46:46.720 offered a personalized view 00:46:46.720 --> 00:46:49.520 of the current stance that we're taking 00:46:49.880 --> 00:46:52.720 so that it gets confirmed again and again and again. 00:46:53.080 --> 00:46:56.480 So I'm not talking about that type of personalization. 00:46:56.840 --> 00:47:01.920 I'm talking about the type of personalization where the interface 00:47:02.480 --> 00:47:06.320 makes it easier for us to participate and addresses 00:47:06.320 --> 00:47:09.720 our specific, very diverse requirements with respect 00:47:10.120 --> 00:47:12.760 to that participation so that 00:47:13.760 --> 00:47:17.480 I've moved away from the term personalization simply 00:47:17.480 --> 00:47:22.760 because I don't want it to be mistaken for the type of personalization 00:47:23.120 --> 00:47:27.080 that cushions us away from diverse perspectives. 00:47:27.080 --> 00:47:32.160 Because certainly we need to be exposed to those that diversity of perspectives, 00:47:33.080 --> 00:47:35.600 and we need to consider 00:47:35.600 --> 00:47:40.160 the diverse stories that people have. 00:47:40.160 --> 00:47:47.080 You know, I think personalization, personalization is 00:47:48.440 --> 00:47:52.480 really part of accessibility in general, 00:47:52.480 --> 00:47:55.560 but there's 00:47:55.560 --> 00:47:58.760 you know, you're talking about a particular kind of personalization, 00:47:59.160 --> 00:48:00.520 but AI personalization. 00:48:00.520 --> 00:48:03.640 I'm going to talk a little bit more in the keynote at the end about an example 00:48:03.640 --> 00:48:06.920 of AI personalization of personalized models 00:48:06.920 --> 00:48:09.440 that are 00:48:12.800 --> 00:48:14.560 providing personalized access 00:48:14.560 --> 00:48:19.040 to digital content, which I think is a good use of personalization. 00:48:19.400 --> 00:48:20.880 Yeah, Yeah. 00:48:21.800 --> 00:48:25.040 So Carve or Convey 00:48:25.520 --> 00:48:30.200 from EDF says thank you Jutta for this important keynote. 00:48:30.200 --> 00:48:34.400 I've seen different toolkits to test and mitigate bias in AI. 00:48:34.400 --> 00:48:39.640 What is your view on them and their usefulness? 00:48:39.640 --> 00:48:42.080 Right. So we've been doing a 00:48:43.160 --> 00:48:44.280 well actually 00:48:44.280 --> 00:48:48.840 as part of a number of our projects, including ODD, which is optimizing 00:48:48.840 --> 00:48:52.520 diversity with disability and our We Count project, 00:48:52.800 --> 00:48:57.400 we've been looking at a variety of A.I. 00:48:57.400 --> 00:49:02.360 ethics, auditing tools, but also we've done sort of the secret shopper 00:49:03.080 --> 00:49:05.960 test of employment tools 00:49:06.320 --> 00:49:11.680 and then seen whether we can detect the the particular biases 00:49:12.000 --> 00:49:16.640 that unwanted biases as is is made clear by AI. 00:49:16.640 --> 00:49:20.480 So I mean bias of course, the tools are intended to be biased. 00:49:21.080 --> 00:49:24.680 And so it's the unwanted bias as a proviso. 00:49:24.960 --> 00:49:27.760 And what we find is that 00:49:28.120 --> 00:49:30.880 they're great at cluster analysis 00:49:31.640 --> 00:49:34.400 and then they they supplement the cluster 00:49:34.400 --> 00:49:39.120 analysis with a number of questions that is 00:49:39.120 --> 00:49:42.880 asked of the implementer of the system. 00:49:43.320 --> 00:49:46.440 So the primary technical key to to 00:49:46.440 --> 00:49:51.440 the tools is determining whether there is unfair 00:49:51.440 --> 00:49:54.560 treatment of one bounded group with another. 00:49:54.560 --> 00:49:57.040 And that works well if you have something like 00:49:57.840 --> 00:50:02.760 determining whether there's discrimination regarding gender or discrimination 00:50:03.440 --> 00:50:08.400 regarding declared race or language or those sorts of things 00:50:08.640 --> 00:50:13.200 which do cluster well, but it doesn't 00:50:13.400 --> 00:50:18.560 none of the tools really detect whether there is discrimination 00:50:19.600 --> 00:50:22.120 based upon disability. 00:50:22.120 --> 00:50:25.160 And and the 00:50:25.160 --> 00:50:27.400 because the the particular 00:50:28.680 --> 00:50:33.760 discriminating characteristics are so diffuse and different 00:50:33.760 --> 00:50:38.960 from person to person, we can't see how it would be possible 00:50:39.320 --> 00:50:42.560 in a litigation perspective 00:50:42.560 --> 00:50:45.080 or in a regulatory perspective to prove 00:50:45.560 --> 00:50:49.080 that you have been discriminated against it. 00:50:49.240 --> 00:50:53.080 It's going to be very, very difficult to come up with that proof 00:50:53.560 --> 00:50:57.440 because the the particular characteristics are 00:50:59.080 --> 00:51:01.320 themselves so entangled and diffuse. 00:51:01.640 --> 00:51:05.600 And so it may not be one particular characteristic associated 00:51:05.600 --> 00:51:08.240 with your disability that you would use to say, 00:51:08.240 --> 00:51:11.480 well, look at here, I'm being discriminated against 00:51:11.480 --> 00:51:16.640 because of this characteristic that relates to my disability. 00:51:16.640 --> 00:51:17.000 Yeah. 00:51:17.000 --> 00:51:20.920 So I think there are a lot of the toolkits. 00:51:21.320 --> 00:51:24.360 Many of the methods in the toolkit are group fairness 00:51:24.840 --> 00:51:27.200 metrics like, like you say, where. 00:51:28.480 --> 00:51:30.920 And that's an important thing to measure. 00:51:30.920 --> 00:51:34.560 It is when we when we do have the ability 00:51:34.560 --> 00:51:39.320 to identify groups and to know for sure who's in which group and which one, 00:51:39.960 --> 00:51:43.520 the boundaries of these groups are always not fuzzy. 00:51:43.520 --> 00:51:46.080 You know, there's the there's 00:51:47.440 --> 00:51:50.080 a deeply embedded assumption that there's 00:51:50.080 --> 00:51:54.000 only two genders, for example, in a lot of the data. 00:51:54.000 --> 00:51:56.000 And many of these tools. 00:51:56.000 --> 00:52:01.480 So they have their problems and this ability exemplifies this. 00:52:01.640 --> 00:52:04.720 The same problem. Yeah, same problems. 00:52:04.920 --> 00:52:09.240 But there are also individual fairness metrics and measures, 00:52:09.600 --> 00:52:13.240 and some of the toolkits include some of these kinds of measures. 00:52:13.240 --> 00:52:16.800 And so instead of asking, is this group as a whole, 00:52:17.000 --> 00:52:21.080 treated equivalently to this other group, they ask 00:52:21.720 --> 00:52:26.200 is a similar are similar individuals treated similarly? 00:52:26.200 --> 00:52:29.320 And so you could imagine with an approach like that, 00:52:29.720 --> 00:52:33.920 if I as an individual with my unique data, 00:52:34.920 --> 00:52:35.520 I could make a 00:52:35.520 --> 00:52:38.000 case that I was discriminated against by 00:52:39.400 --> 00:52:41.880 creating another person 00:52:42.200 --> 00:52:45.080 who was similar to me in the respects 00:52:45.080 --> 00:52:47.440 that are important for this job. 00:52:47.440 --> 00:52:49.240 Mm hmm. Yeah. 00:52:49.720 --> 00:52:53.240 And. And see what kind of result they got compared to my result. 00:52:53.240 --> 00:52:56.480 And that would be, you know, a way to measure 00:52:56.480 --> 00:52:59.800 individual fairness and and build up a case. 00:53:00.400 --> 00:53:01.640 Yes. Yeah. 00:53:01.640 --> 00:53:05.480 Unfortunately, there's not that many schools that that currently do that. 00:53:05.480 --> 00:53:08.280 And the certification systems that currently exist 00:53:08.640 --> 00:53:12.720 are not implementing those so that there is much to work on there. 00:53:13.400 --> 00:53:14.000 Yeah. 00:53:14.000 --> 00:53:17.240 Think it's sort of more of a case, a 00:53:21.000 --> 00:53:22.160 case by case basis, 00:53:22.160 --> 00:53:25.320 but this particular job to me 00:53:25.320 --> 00:53:29.320 so it's not so easy to make a blanket statement about it, 00:53:29.320 --> 00:53:31.960 but I think it's not impossible to 00:53:32.840 --> 00:53:38.000 assess. 00:53:38.000 --> 00:53:39.400 Okay, so 00:53:43.000 --> 00:53:44.480 do we have time for one more? 00:53:44.480 --> 00:53:44.920 How long? 00:53:44.920 --> 00:53:47.880 How much longer do we have another three minute? 00:53:48.920 --> 00:53:51.600 Well, you have almost 10 minutes more, 00:53:51.600 --> 00:53:54.000 so you can definitely take one more. 00:53:55.160 --> 00:53:59.920 Awesome, great. Um, 00:54:01.800 --> 00:54:11.240 let's see. 00:54:11.240 --> 00:54:15.680 So Fabian Berger 00:54:16.640 --> 00:54:18.920 says, I feel that AI 00:54:18.920 --> 00:54:23.560 but before it was KPIs or else are 00:54:23.640 --> 00:54:28.120 are searched by managers to justify their decisions or run away 00:54:28.440 --> 00:54:30.920 from the responsibility of their decisions. 00:54:32.480 --> 00:54:34.520 It follows a need for them 00:54:34.800 --> 00:54:37.360 but with a wrong incomplete answer. 00:54:37.960 --> 00:54:42.200 Do you agree? Yes. 00:54:42.440 --> 00:54:46.080 I mean, I think that the issue 00:54:46.080 --> 00:54:50.080 and I was trying to make that point, but possibly not well enough 00:54:50.960 --> 00:54:54.560 that AI is doing much of what 00:54:54.560 --> 00:54:57.920 we've done before, but 00:54:57.920 --> 00:55:00.200 it's amplifying, accelerating 00:55:00.200 --> 00:55:03.240 and automating those things. 00:55:03.560 --> 00:55:06.320 And certainly AI can be used to 00:55:06.400 --> 00:55:10.640 for confirmation bias to find the specific 00:55:11.760 --> 00:55:14.280 justification for what it is 00:55:14.280 --> 00:55:18.400 that we need to justify whether it's something good or something bad. 00:55:18.400 --> 00:55:22.680 So the a lot of the the harms of AI 00:55:22.720 --> 00:55:25.040 already existed 00:55:26.360 --> 00:55:29.240 because of course AI is learning 00:55:29.240 --> 00:55:31.720 our past practices and our data, 00:55:32.520 --> 00:55:35.240 but because it 00:55:35.520 --> 00:55:38.720 I guess I've often used the analogy of a power tool 00:55:39.200 --> 00:55:41.280 before it was this 00:55:42.360 --> 00:55:47.240 practice that was not that we did manually. 00:55:47.240 --> 00:55:51.440 And so there was an opportunity to make exceptions, 00:55:51.440 --> 00:55:55.960 to reconsider, you know, is this actually what we're we're doing? 00:55:56.360 --> 00:55:58.640 And to, to do something different? 00:55:58.640 --> 00:56:01.720 But with the with a power tool, it's 00:56:02.480 --> 00:56:06.920 it becomes this much more impactful thing. 00:56:07.400 --> 00:56:10.680 And there's less opportunity to 00:56:10.720 --> 00:56:15.560 to craft the approach that we take. 00:56:15.560 --> 00:56:18.680 Yeah, I think that's why it's really important to 00:56:19.520 --> 00:56:21.680 to try to design for outliers 00:56:21.840 --> 00:56:24.800 and consider outliers in in and again 00:56:24.800 --> 00:56:27.480 I come back to this point of a system 00:56:27.800 --> 00:56:32.360 that the system as a whole that includes AI if we can't 00:56:34.040 --> 00:56:35.440 guarantee that the AI 00:56:35.440 --> 00:56:38.440 itself is going to give us the 00:56:38.440 --> 00:56:42.320 characteristics that we want, then we need to design around that 00:56:42.800 --> 00:56:45.680 and and be mindful of that 00:56:45.680 --> 00:56:51.040 while we're designing. 00:56:51.040 --> 00:56:57.600 There's also, of course, the opportunity to try to clean up our data in general 00:56:58.000 --> 00:57:01.960 is there are in situations where we can identify 00:57:03.160 --> 00:57:05.640 problems with the data, we should certainly tackle 00:57:05.640 --> 00:57:07.440 that or imbalances in the data. 00:57:07.440 --> 00:57:08.720 We should certainly tackle that. 00:57:08.720 --> 00:57:11.640 That's one you know, one other step that 00:57:11.760 --> 00:57:15.200 I think there's many steps to fairness and 00:57:16.160 --> 00:57:19.000 and to ethical application of AI. 00:57:19.520 --> 00:57:21.560 And no one step 00:57:22.480 --> 00:57:25.680 is a magic solution to all of that. 00:57:25.680 --> 00:57:27.440 But if we 00:57:27.800 --> 00:57:32.280 stay aware of the risks, make sure we're talking 00:57:32.280 --> 00:57:37.200 to the right people and involving them, then I think we can we can at least 00:57:38.360 --> 00:57:40.440 mitigate problems and know 00:57:40.440 --> 00:57:42.800 the limits of the technologies that we're using 00:57:44.680 --> 00:57:47.720 better. 00:57:47.720 --> 00:57:51.440 Yeah, I've been looking at some of the other questions that have come in 00:57:51.840 --> 00:57:57.800 and one of the discussions was about the Gaussian curve and or the Gaussian center. 00:57:57.800 --> 00:58:00.080 And one thing I think that 00:58:01.120 --> 00:58:03.840 one point that I may not have made as clearly 00:58:03.840 --> 00:58:07.640 is that in fact the the myth 00:58:07.640 --> 00:58:11.160 that we need to have a single answer 00:58:11.520 --> 00:58:17.000 at the very middle of the Gaussian curve, which of course matches our notion 00:58:17.000 --> 00:58:21.560 of majority rules or as the way to decide amongst 00:58:21.560 --> 00:58:23.680 difficult decisions, it 00:58:24.880 --> 00:58:28.200 an alternative to that is to address 00:58:28.200 --> 00:58:31.160 the very, very diverse edges 00:58:32.280 --> 00:58:34.520 initially and to prioritize those. 00:58:34.520 --> 00:58:39.560 Because what what then happens is it gives us room 00:58:40.560 --> 00:58:43.240 for to change. 00:58:43.640 --> 00:58:45.800 It helps us address the uncertainty. 00:58:45.800 --> 00:58:49.040 It makes the whole design or 00:58:49.280 --> 00:58:52.480 decision or options that are available 00:58:54.440 --> 00:58:56.000 much more 00:58:56.000 --> 00:58:59.480 generous and therefore prepares us better 00:58:59.480 --> 00:59:03.680 for the vulnerabilities that we're going to experience in the future. 00:59:03.680 --> 00:59:04.840 So that, 00:59:05.520 --> 00:59:08.440 of course I'm an academic, and to say that 00:59:08.800 --> 00:59:11.640 that statistical reasoning and evidence 00:59:11.960 --> 00:59:14.760 through scientific methods is at 00:59:14.760 --> 00:59:19.320 fault is is a fairly dangerous thing to say, especially during a time 00:59:19.720 --> 00:59:23.440 when truth is so much under attack and 00:59:24.000 --> 00:59:29.000 but I think what we need to do is not reduce truth to statistical reasoning, 00:59:29.680 --> 00:59:32.120 but to acknowledge that there are 00:59:32.840 --> 00:59:37.880 a variety of perspectives, untruth, and that we need to come up with one 00:59:37.880 --> 00:59:43.040 that that addresses the people that we're currently excluding 00:59:43.040 --> 00:59:58.400 in our notions of truth. 00:59:58.400 --> 01:00:01.480 Yeah, So there's 2 minutes left. 01:00:01.720 --> 01:00:05.080 I, I think now 01:00:05.880 --> 01:00:07.800 maybe we could squeeze in one more 01:00:07.800 --> 01:00:12.640 question here. 01:00:12.640 --> 01:00:14.280 Jen Benjamin 01:00:15.440 --> 01:00:17.040 asks, Do we think 01:00:17.040 --> 01:00:23.120 that AI and big companies driving research on it can be problematic towards 01:00:23.120 --> 01:00:27.240 societal issues which don't necessarily give the highest revenue? 01:00:29.040 --> 01:00:31.440 And if so, how can it be faced? 01:00:33.400 --> 01:00:35.920 Yeah, that's a huge question. 01:00:36.320 --> 01:00:37.680 I think even government 01:00:37.680 --> 01:00:43.400 efforts are facing in their decision making on profit 01:00:43.400 --> 01:00:45.920 and economic progress and 01:00:47.800 --> 01:00:50.240 and impact measures that are 01:00:51.000 --> 01:00:55.760 so that I think one of the things we need to abandon 01:00:56.160 --> 01:01:00.960 is this idea that a solution needs to be formulaic 01:01:00.960 --> 01:01:04.480 and that we need to scale it by a formulaic replication. 01:01:05.120 --> 01:01:08.000 And we need to recognize that 01:01:08.600 --> 01:01:11.480 that there's a different form of scaling 01:01:11.920 --> 01:01:14.760 that is by diversification, 01:01:15.200 --> 01:01:18.560 and that we need to contextually apply things. 01:01:19.080 --> 01:01:25.680 And I mean, that's one of the lessons of Indigenous and indigenous cultures 01:01:27.200 --> 01:01:29.280 that the 01:01:29.280 --> 01:01:34.040 what, what has been labeled as colonialist is, is actually what many governments 01:01:34.040 --> 01:01:40.520 are still implementing, even in things like social entrepreneurship. 01:01:41.000 --> 01:01:44.680 But yes, big companies, of course, are driven by profit and 01:01:46.160 --> 01:01:49.160 is that the best approach 01:01:49.600 --> 01:01:52.680 to achieve the common good? 01:01:52.680 --> 01:01:54.840 That's a huge question. 01:01:54.840 --> 01:01:55.520 I think. 01:01:55.520 --> 01:01:59.320 And I think this would be a great one to come back to tomorrow, actually. 01:01:59.320 --> 01:02:00.400 Yeah, exactly. 01:02:00.400 --> 01:02:03.200 I mean, the symposium let's let's come back to that one 01:02:03.200 --> 01:02:05.680 because I see that we're out of time right now. 01:02:05.680 --> 01:02:08.080 But thank you very much. 01:02:08.320 --> 01:02:09.560 Thank you.