We must have a legal framework dealing with the “full range” of issues around artificial intelligence, including algorithmic bias, data mining, using data without the permission of the data owner, and broad principles for the same must be set out “a priori”, former TRAI chairman R.S. Sharma said on Wednesday. In order to deal with the onset of artificial intelligence, India must also have a robust data protection legislation; “the sooner it comes the better it is”, Sharma said, while speaking at the Indian government’s Raise 2020 Summit (Responsible AI for Social Empowerment, 2020).
When asked about the ways in which AI can be regulated, Sharma said that regulations should be largely “reactive”, that is “you don’t regulate things unless there is a market failure”. He cautioned against doing “anticipatory regulation”, warning that it might prevent innovation, which is “not the right way to regulate technologies”.
Instead, Sharma continued, India can lay down “principle based” regulations which don’t really impact innovation, or fair play. “For example, we have DEPA [Data Empowerment and Protection Architecture], which ensures that a data owner’s data is used for their own betterment. I think we should make such architectures,” Sharma said. DEPA seeks to break data silos and monopolies so that fintech and healthtech companies can compete on basis of design, analytics and value creation rather than data access.
‘Be wary of transnational large corporations’
Aside from the inherent issues of AI as a technology, Sharma also cautioned against the people and few companies who develop those algorithms. “So what is happening is that the people who are making these algorithms, make them in a way that their objective function is to direct people in a certain direction, create biases in their minds, then it works in a subconscious level which you can’t figure out is affecting you,” Sharma said.
“These solutions are concentrated in the hands of a few large corporates who are not really responsible for any government, they are larger than the government themselves. Our laws and regulations are national, but these companies are transnational,” Sharma remarked. He continued: “If you remember, there used to be a number of social media platforms in the past, and now there is nobody, but only Facebook”, and added that “we must ensure that our data empowers us, and not just serve the purpose of those who are mining that data”.
“Some of the principles which we must adopt for [safeguards against] this are the principles which we have adopted in many other technologies in our country. For instance, we have built open systems based on open APIs, interoperability, and frugality, among others,” Sharma said and proposed the following ideas to deal against the power that large companies wield:
- Decentralisation: Internet is a democratic structure, no one owns it, it is operated collectively. If we can put in place Similar architecture if we can put in place.
- Open algorithms: Opening up algorithms. We must know what is the source code, how it impacts us, and if that could become a general norm.
‘AI can also be misused’
“I agree that AI has a huge amount of capability, and while these capabilities can be used for the betterment of human kind, it can also be misused. I have always believed that technology doesn’t have any ethical part into it. It is the ethics which we introduce into it, technology otherwise is agnostic to any of these concepts,” Sharma said. He said that AI brings in “intelligence into the technology”, and provides the power of decision making to the technology itself, therefore it becomes much more capable. “So the attributes of AI, are extremely powerful, and raise concerns whether AI will someday overtake humans or not”.
AI has overtaken many of the functions and there are dangers of algorithmic bias, he said. “There are dangers of lot of expert jobs being taken away by AI and machines. But if we ensure that we are not just becoming consumers of these technologies, and do not actively participate from the beginning itself to preserve our own interests, then it’ll be very good”.
Drawing inspiration from Aadhaar and the Unified Payment Interface (UPI), Sharma said that India has the capability to develop unique solutions without necessarily copying what the rest of the world is doing. “Let’s understand that India has developed solutions which are unique, and not really just what the west has done. We have a digital identity [Aadhaar], a payment system [UPI] which is so scalable that we are doing billions and billions of transactions per month. I think we have the tools and knowledge to build systems on top of these systems which will serve our purpose,” he said.
- Privacy of users should be protected from private contracts: RBI executive director
- Data a ‘national resource’, confident the govt will introduce ‘sound’ data regulation framework to protect it: Mukesh Ambani
- UIDAI, NPCI piloting face authentication for Aadhaar
- Summary: Data Empowerment and Protection Architecture (DEPA)