AI innovation: Why all players do not have a fair shot

As interest around artificial intelligence grows, tech companies are trying to outdo each other in the development of this emerging technology. But some players have an uneven advantage over others when it comes to development of AI projects, highlights Bhaskar Roy, Client Partner for Asia-Pacific at Fractal Analytics, India’s first AI unicorn.

What is driving AI innovation?

Roy explains that there are two main factors driving AI innovation– data and powerful GPUs. He highlights that some players have an unfair advantage when it comes to both these factors.

In a conversation with Business Today, Roy said, “What is driving this advancement in AI is essentially twofold. One, the highly powerful GPU chipsets and two, the data that underlies the model.”

Big tech's advantage

Speaking of access to data, the Client Partner at the unicorn AI company explains that there are multiple ways this data can be obtained, “For example, if I were to start my own company today, I have three ways in which I can access new data to actually build a product. First of course are open sources that are available. Second, I can go and buy data from some of these large corporations. Third is to go and do primary research and collect data on my own, which is extremely expensive.”

Roy notes that the fact that many of the major big tech players have tons of data on their users, gives them an unfair advantage.

"Big tech companies have collected the data for a fraction of the cost, the data which newer players would have to pay them for to build their own AI models," he explains.

Semiconductor politics

Not just data, Roy adds, some geographical locations are also at a disadvantage when it comes to GPU chipsets.

“Semiconductors are sometimes monopolised by certain countries, either by way of producing them, or even by way of sanctions,” he adds.

It is worth noting that semiconductors have become a tool of geopolitical warfare in recent times. A recent example of this is US’ sanctions and restrictions on export of NVIDIA and AMD chips to China and Middle Eastern countries.

How can there be equitable access?

Fractal’s Roy explains that one way to address the equitable access of data is by establishment of data sharing sandboxes

“One way to address this is to establish data-sharing sandboxes where data could be anonymised and made available for the larger public good, so that all the players get a fair shot at new developments in the field of AI,” he explains.

He also adds that a global consensus is required to keep anti competitive practices and sanctions at bay. “All the economies need to come together and understand that innovation is promoted and everyone has a fair shot at building the next advancements in AI.” 

Global AI regulations

Business Today had reported that ethical development of AI and its regulation were discussed at the G20 Summit in New Delhi last weekend. Moreover, a highly placed source told Business Today that AI regulation and its ethical use would be an important point of discussion at G20 Brazil, the G20 Leaders have concluded.