What Do You Do with a Problem Like AI?
Download as PPTX, PDF2 likes424 views
The document summarizes key topics around artificial intelligence, autonomous vehicles, and regulation. It discusses what AI is, including narrow and general AI, as well as machine learning approaches. It also covers applications of AI like autonomous vehicles and issues around regulation, fairness, and safety. New regulators like an EU agency for robotics and AI are proposed to help govern emerging technologies.
![Why are algorithms important to society, governance,
innovation?
Predictive profiling – and hence manipulation - of persons
Targeted ads on Google/social networks, etc; manipulation of personalised newsfeeds
Price , service discrimination;
Criminal/terrorist profiling; -> pre-crime?
Future health, obesity, Alzheimers risks
Non-personal predictions of what is important/significant/popular/profitable;
eg “trending topics” on Twitter;
Google News; top Google results on search by keywords; top FB links to “fake news”?
Automated high frequency trading on stock exchanges;
Recommendations on Netflix/Amazon etc
Filtering online of unwanted content
Spam algorithms, Google Hell (anti SEO)
Over, under block? Twitter UK anti-women trolling cases summer 2013: ACPO “They [Twitter] are ingenious people, it can't be beyond their
wit to stop these crimes”
“Real world” as well as online effects: Algorithms to instruct robots on how to behave adaptively when circumstances
change from original programming; driverless cars liability?
See *Kohl (2013) 12 IJLIT 187.](https://image.slidesharecdn.com/airobotsandthelaw-161222163608/85/What-Do-You-Do-with-a-Problem-Like-AI-5-320.jpg)
![EU legal remedies? Data Protection
Directive – transparency of algorithms
Art 12: "every data subject [has] the right to obtain from the controller..
- knowledge of the logic involved in any automatic processing of data c at least in
the case of the automated decisions referred to in Article 15 (1)“
Art 15(1) : every person has the right "not to be subject to a decision which
produces legal effects concerning him or significantly affects him and which is
based solely on automated processing of data iintended to evaluate certain
personal aspects relating to him, such as his performance at work,
creditworthiness, reliability, conduct, etc.“
Rec 41: "any person must be able to exercise the right of access to data relating to
him which are being processed, in order to verify in particular the accuracy of the
data and the lawfulness of the processing“
..” this right must not adversely affect trade secrets or intellectual property and in
particular the copyright protecting the software”](https://image.slidesharecdn.com/airobotsandthelaw-161222163608/85/What-Do-You-Do-with-a-Problem-Like-AI-16-320.jpg)
![Draft DP Regulation (Jan 16)
New Art 15: Rts of access
Right to obtain where personal data is being processed..
“(h) the existence of automated decision making including profiling [see art
20] .. And at least in those cases, meaningful information about the logic
involved, as well as the significance and envisaged consequences of such
processing..”
*Rec 51: “This right should not adversely affect the rights and freedoms of
others, including trade secrets or intellectual property…However, the result
of these considerations should not be that all information is refused to the
data subject…”](https://image.slidesharecdn.com/airobotsandthelaw-161222163608/85/What-Do-You-Do-with-a-Problem-Like-AI-17-320.jpg)
What Do You Do with a Problem Like AI?
- 1. AI, autonomous vehicles and the Law Lilian Edwards November 2016
- 2. Preparing for the Future of AI, US, Oct 2016 What is AI? Application of AI to the Public Good AI and Regulation Research and Workforce Economic (employment) Impact of AI Fairness, safety and Governance Global Considerations and Security Lethal autonomous weapons
- 3. What is AI? “Although the boundaries of AI can be uncertain and have tended to shift over time, what is important is that a core objective of AI research and applications over the years has been to automate or replicate intelligent behaviour” Narrow AI v General AI; the path to the singularity? “long term concerns about superintelligent General AI should have little impact on current policy” Machine Learning vs expert systems Experts come up with rules, AI systems implement them Vs statistical methods used to find a “decision procedure” that “works well in practice” Adv of ML : can be used where difficult or impossible to come up with explicit rules eg fraudulent logins, reversing a lorry Disadv: the system generated may not easily exlain how or why it comes to the decisions it does = “black box”
- 4. How ML works Needs large amounts of data (“big data”); algorithm; processing power. Improvements in all since 80s. Which has ben most important? Creating a ML system DATA. Divide into training set and test set. Create a model of how system works ie a set of possible mathematical rules Adjust the parameters of the model (creates millions of options) Define a successful result (objective function) Using training set of successful decisions, train system to get from data to outcome in most efficient way Test it works using test set Hope it can now generalise ie successfully apply induced rules to new examples not in training set. You have created the ALGORITHM. Essentially a process of statistical correlation – often not clear why input data X lead to result Y cf an expert system which had a human-produced underlying rationale DEEP LEARNING: many layers of “neurons” are used in training the system – resembling how brain processes and creates knowledge
- 5. Why are algorithms important to society, governance, innovation? Predictive profiling – and hence manipulation - of persons Targeted ads on Google/social networks, etc; manipulation of personalised newsfeeds Price , service discrimination; Criminal/terrorist profiling; -> pre-crime? Future health, obesity, Alzheimers risks Non-personal predictions of what is important/significant/popular/profitable; eg “trending topics” on Twitter; Google News; top Google results on search by keywords; top FB links to “fake news”? Automated high frequency trading on stock exchanges; Recommendations on Netflix/Amazon etc Filtering online of unwanted content Spam algorithms, Google Hell (anti SEO) Over, under block? Twitter UK anti-women trolling cases summer 2013: ACPO “They [Twitter] are ingenious people, it can't be beyond their wit to stop these crimes” “Real world” as well as online effects: Algorithms to instruct robots on how to behave adaptively when circumstances change from original programming; driverless cars liability? See *Kohl (2013) 12 IJLIT 187.
- 7. “Latanya Sweeney, Arrested? 1) Enter name and state 2) Access full background. Checks instantly. www.instantcheckmate.com”
- 8. Legal/regulatory issues EXAMPLE: Autonomous cars Huge advs – reducing road deaths; access fo diabled’ access for the rural economy; environmental advantafges Safety and security issues crucial to autonomous cars uptake Degrees of autonomy – we are already part the way (4 levels) “the goals and structures of existing regulation are sufficient and commentators called for existing regulation to be adapted as necessary” Dealing with liability of operators; dealing with exceptions. Unusual circumstances (“trolley problem”) Role for existing agencies eg FAA, CAA in UK ; need for them to have access to tech experts Most countries passing laws to allow trials on (restricted) public roads
- 10. Regulatory problems – 1. Fairness 1. Worries re quality of training and testing data; justice; fairness, esp in criminal justice eg “Risk prediction” tools used for criminal sentencing or bail decisions – data may be partial or poor quality -> racial prejudice (Angwin, Pro Publica, May 2016) Eg hiring systems – used to screen job applicants – may reinforce bias already in system by hiring more who like look like current workforce Remedies? Transparency: systems should have to be able to demonstrate fairness if questioned – accountability akin to judicial review? “Having an interpretable model helps Better testing to weed out bad results: eg gorilla face recognition eg “in the wild” testing eg open source scrutiny Ethical training for AI researchers
- 11. Regulatory problems – Safety and Security Transition from safety of lab to unpredictable “in the wild” – racist chatbot? Open to remote hacking from third parties without obvious detection Especially worrying for Internet of Things/ real world applications – eg Tesla experience Need for a “matue field of AI safety engineering” Also need for a database of outcomes so risks can be actuarialised and insurance provided – Modern Transport Bill 2016 (UK) ties to kickstart this Global cooperation needed for AI cybersecurity (pre Trump!) Passing mention of “privacy and security” (p 36 US report)
- 14. Effect on society and employment? AI needs dramatic growth in skilled workfoce Current workforce notably undiverse both in colour and gender However also likely to lose jobs via automation – primarily blue collar jobs Alternately middle class jobs may become more efficient by augmentation/ co working with AI Heightened inequality Lawyers? “AI Judges” a long way off..?
- 15. “In the course of developing the programme the team found that judgments of the European court of human rights depends more on non-legal facts than purely legal arguments.” “The algorithm examined English language data sets for 584 cases relating to torture and degrading treatment, fair trials and privacy.” “The AI “judge” has reached the same verdicts as judges at the European court of human rights in almost four in five cases involving torture, degrading treatment and privacy.”
- 16. EU legal remedies? Data Protection Directive – transparency of algorithms Art 12: "every data subject [has] the right to obtain from the controller.. - knowledge of the logic involved in any automatic processing of data c at least in the case of the automated decisions referred to in Article 15 (1)“ Art 15(1) : every person has the right "not to be subject to a decision which produces legal effects concerning him or significantly affects him and which is based solely on automated processing of data iintended to evaluate certain personal aspects relating to him, such as his performance at work, creditworthiness, reliability, conduct, etc.“ Rec 41: "any person must be able to exercise the right of access to data relating to him which are being processed, in order to verify in particular the accuracy of the data and the lawfulness of the processing“ ..” this right must not adversely affect trade secrets or intellectual property and in particular the copyright protecting the software”
- 17. Draft DP Regulation (Jan 16) New Art 15: Rts of access Right to obtain where personal data is being processed.. “(h) the existence of automated decision making including profiling [see art 20] .. And at least in those cases, meaningful information about the logic involved, as well as the significance and envisaged consequences of such processing..” *Rec 51: “This right should not adversely affect the rights and freedoms of others, including trade secrets or intellectual property…However, the result of these considerations should not be that all information is refused to the data subject…”
- 18. New regulators? UK parliament Sci/Tech Committee report on robots and AI, 2016 “Our inquiry has illuminated many of the key ethical issues requiring serious consideration— verification and validation, decision-making transparency, minimising bias, increasing accountability, privacy and safety.” “We recommend that a standing Commission on Artificial Intelligence be established, based at the Alan Turing Institute, to examine the social, ethical and legal implications of recent and potential developments in AI. It should focus on establishing principles to govern the development and application of AI techniques, as well as advising the Government of any regulation required on limits to its progression. It will need to be closely coordinated with the work of the Council of Data Ethics which the Government is currently setting up following the recommendation made in our Big Data Dilemma report.”
- 19. EU parliament draft report on robotics, 2016 “Calls for the creation of a European Agency for robotics and artificial intelligence in order to provide the technical, ethical and regulatory expertise needed to support the relevant public actors, at both EU and Member State level, in their efforts to ensure a timely and well-informed response to the new opportunities and challenges arising from the technological development of robotics; a system of registration of advanced robots should be introduced, based on the criteria established for the classification of robots. The system of registration and the register should be Union-wide, covering the internal market, and should be managed by an EU Agency for Robotics and Artificial Intelligence
- 20. “Code”/ self regulatory remedies? “Those clamoring for Facebook to fix its fake news problem should be careful what they wish for. They might find in a few years that the fake news is gone— but the filter bubbles, the perverse incentives, and Facebook’s pretense to algorithmic neutrality remain.” Slate Nov 16, 2016