The rise of AI has sparked a range of anxieties: about job loss and societal upheaval, about the technology’s toll on the environment, about whether some future form of AI could pose an existential threat to humanity itself.
Those aren’t the main things that keep Ashish Vaswani up at night, though. Instead, he's concerned that the current AI boom may have become a threat to AI itself.
As the financial stakes of AI have increased, Vaswani sees commercialism distorting the scientific field that fuels the technology. Research has migrated from a wide range of established companies, startups and academic institutions to a handful of deep-pocketed corporate labs including Google DeepMind, Microsoft, Meta and OpenAI.
Operating under immense pressure to justify tens of billions of dollars in investment, those labs have dedicated themselves largely to squeezing improvements out of the same transformer-based models that have been so impressive to date, hoarding their insights from one another in the hope of gaining an advantage in the market.
This dynamic, Vaswani fears, runs the risk of blinding the entire field to the most relevant applications while also reducing the chances of finding alternative paths that lead to the next big breakthrough.
Read more | BLOOMBERG