The #AI task force for Economic Transformation in #India (set up by the Government’s Ministry of Commerce and Industry) released its report last month. Some background (B) and key takeaways (T). 1/n
B1. The vision guiding this report is “AI as a socio economic problem solver at large scale rather than only a booster of economic growth”. 2/n
B2. It seeks to address the role of the Government in the field, answer how #AI can improve quality of life, and identify sectors for AI growth. 3/n
B3. The report identifies 10 important domains of#AI relevance to #India, including healthcare, manufacturing, national security, FinTech, Retail, etc. 4/n
T1.To a large extent, it buys into the #AI hype. Timeframe for consideration in 5 years, and the report finds that it is realistic to expect human level or better performance in a variety of #cognitive tasks, including #perception. 5/n
T2.Sectoral analysis are problematic. For instance, opportunities for #FinTech include lending & predictive investment, but challenges find no mention of discrimination, problems with credit scoring. 6/n
T3.In fact, the only challenges are scale and #innovation, anticipation of market demand. Shouldn’t we be considering definitions of what constitutes #fairness? Discuss the issue of demographic parity (to borrow one definition, atleast?) 7/n
T4.Another sector is #NationalSecurity. Report states that #AI can be “usefully deployed” in areas like #autonomous #surveillance. No mention of civil liberties, #privacy, #data protection, etc. 8/n
T5.The only national security challenges are cost, tech availability and systems management capability deficit. 9/n #AI #India
T6. In fact, the only place where #discrimination, #data, #privacy, #accountability are discussed in more than a bulleted list is in the *retail* sector. 10/n
T7. In discussing enablers for the future of #AI, the report states, “positive social attitudes to machines and trust in autonomous systems”. This is dangerous, and its exactly this blind faith in #tech that we need to become more critical of. 11/n
T8. Positive attitudes & trust in autonomous systems cannot, should not be *declared*. It needs to be achieved. 12/n #AI #India
T9. There is a section that discusses responsible use of #AI. Blanket statements calling for #explainability, safety etc are not useful. Eg: Limits, requirements, and dangers vary greatly between national security and retails. 13/n
T10. The recommendations include setting up a national #AI Mission with proposed work of establishing networks, projects, competitions, talent conferences, setting up centers of excellence, building standards, etc. 14/n
T11. No mention of the need to consider #regulation. Regardless of the outcome of that conversation, its crucial to discuss. #AI #India 15/n
T12. Report also states that govt policies need to be framed around corporate stakeholders, educational & legal institutions as “key enablers of #AI technologies. No mention of #government use of AI. 16/n
T13. The report is largely disappointing. It misses the opportunity at looking at critical issues around maximising the potential of #AI, *without eroding fundamental rights*. 17/n
T14. The lack of discussion around the issue of social-ethical-legal questions is not surprising given the composition. The TF is made up of six academicians, five bureaucrats, two independent consultants, and two representatives of Indian IT companies. 18/n
T15. Some individuals I deeply respect are on the TF, but the fact that an important stakeholder was not invited to the table, is glaringly obvious. 19/n
T16. Also, 2 women. Sigh. 20/n
T17. There’s also a whole section on #Aadhaar, but maybe that deserves its own journal article (or 5). #AI #India 21/n
T18. This report is particularly disappointing because it doesn’t actually address issues that are unique, specific, and crucial to #India: rather it buys into the “#AI race” mindset without asking which direction we should run in. 22/22.
For those interested in #AI, India, and/or country specific perspectives. cc: @pranesh @neintara @sunil_abraham @ambersinha07 @mayameme @carlynyst @kncukier @C__Cath @F_Kaltheuner @madhumachi @rcalo @MalloryKnodel @m_c_elish
Cheers!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
