Scientists and tech experts are warning of the dangers of funneling billions into training sophisticated AI models and calling for a more cautious approach. -- BY BEATA SOCHA
A recent open letter to the tech industry from scientists and tech doyens (as well as one Twitter CEO), calls for artificial intelligence (AI) companies to put a break on their AI arms race. They say that if we continue funneling billions of dollars into training more sophisticated models, we may simply miss the mark when we develop a model beyond a narrow, specialist AI. A so-called general AI may be able to determine its own goals and means of achieving them. And could simply turn on us, its creators. Scary, but we’ve seen too many post-apocalyptic movies and TV shows for vague threats to have any real impact.
More tangible, short-term challenges they list include uncontrolled job automation (and a huge wave of unemployment as a result), AI bots spreading misinformation on a mass scale, and rampant fraud. We’ve already seen the first malicious uses of AI, such as a very recent attempt to extort ransom from a father by having AI mimic a distressed call from his daughter. Chilling, yes.
The points about a cautious approach to AI are valid, and ideally, we could all agree to take a step back, reassess and reevaluate our course. Ideally. The thing is that historically there has been little room for idealists in the business world. Granted, some tech visionaries could be considered idealistic. Or maybe that’s the narrative they managed to spin years or decades after joining the exclusive club of ultra-rich philanthropists who cure malaria.
1. Revolutions generally don’t come gradually. They happen fast and disrupt industries (streaming services, social media, smartphones, search engines). Unless they fail to capture the market and fizzle out (NFTs, beta max, blu-ray) or crash spectacularly (dot coms, crypto). But once a tech revolution is underway, it is virtually impossible to stop it. The cat’s out of the bag. We now have LLMs (Large Language Models), such as ChatGPT, which can do remarkable things, for better or worse.
2. There is no economic incentive to let your competitors get to market first, especially if you’re trying to capture a burgeoning new segment whose value is yet to be determined. And corporations have a very clear fiduciary responsibility and a well-defined mandate: make your shareholders happy.
Recent moves by Disney and other tech and media giants leave little room for doubt about where the money is. As Fast Company reports, Disney, following in the footsteps of Meta, scrapped its metaverse project and is laying off thousands, gearing up instead for the AI chase.
