Does funding for AI need a sober approach?

Earlier this month, Amazon hit the headlines for rolling back ‘Just Check Out’. This technology was introduced in 2016 at Amazon Fresh, its physical stores, in the US. Customers could simply load their purchases into a cart and skip check out if they had scanned a QR code on entering. This seemed outstanding because everything appeared automated. Over time, flaws started to emerge. The wait time for receipts got longer; and once in a while people got the wrong receipts. It turned out there was no technology at work, just a thousand-odd Indian employees monitoring people and their purchases on cameras. It was just another way of outsourcing checkouts to cheaper cost centres—like IT some years ago.

Amazon has come clean since and has implemented a more sophisticated technology. But a Bengaluru-based public policy professional with a ring-side view of how things work in India wasn’t surprised one bit. “AI is like magic pixie dust. One robotics advisory firm I’ve been in touch with said they use AI but when I talked to someone within that organization they laughed and said all they have is spreadsheets.” He declined to come on the record because of his affiliations.

Now, how do we interpret the Amazon story and what the public policy professional is a witness to? The roots of AI can be traced back to the 1940s with visionaries like Alan Turing pondering over the possibility of machines exhibiting intelligent behavi o ur . Early milestones included the development of neural networks, inspired by the human brain’s structure and function. However, these early forays were limited by computational power and the nascent understanding of how to train these networks.

The decades that followed were marked by periods of excitement and disillusionment. The 1960s saw a surge in AI research funding, only to be followed by the “AI Winter” of the 1970s and 80s, where limitations of technology and unrealistic expectations led to a significant decline i n research. However, the embers of AI never truly died. Advancements in computer science, coupled with the explosion of data available in the digital age revitalized the field in the late 20th century.

The key turning point came with the emergence of machine learning (ML), a subfield of AI that allows algorithms to “learn” from data without explicit programming. Techniques like deep learning, using artificial neural networks with multiple layers, further boosted capabilities. These advancements transformed AI from a theoretical concept to a practical tool. Also, there is now greater accessibility. Most of us can use CHATGPT, Gemini, Grok and applications that include chatbots and virtual assistants with ease. How things are now can be likened to the early days of the internet when basic websites were hailed as revolutionary. Human-level AI is still at a distant horizon. But this is not to suggest the current generation of tools do not hold the potential to revolutionize various fields from healthcare to finance. Just when that leap will happen is something no one knows yet.

That is why the Bengalurubased public policy professional suggests a sober approach to AI. “Every technology goes through a hype cycle. The charlatans leverage investors’ fear of missing out (FOMO) and raise money using the buzzword. Companies are eager to brand themselves as bastions of AI innovation, he says, from self-driving cars to personalized recommendations. But beneath the glossy veneer of buzzwords lies a complex reality. Much of what’s touted as AI isn’t true intelligence but a clever application of established technologies.”

Investor Krishna Jha however is optimistic. Jha who founded the technology start-up Itfinity which got sold to Onmobile, a spinoff from Infosys, says calls for too much caution on AI investments are unnecessary. “The limitations of AI are known at this stage. At the same time, the leaps in technology are unprecedented and in a few years, we will have more reliable use cases. As it is now, and projected, technology is indistinguishable from magic!”

MUCH OF WHAT’S TOUTED AS AI IS JUST A CLEVER APPLICATION OF ESTABLISHED TECHNOLOGIES, SAY EXPERTS