In last week’s Lunch with the FT, George Hammond discussed the hype of generative AI over sushi with Emily Bender (the academic who authored the Stochastic parrots paper and coauthored The AI Con (2024)).
So – where will they land at the end of lunch? Does the rise of AI technology portend a dangerous dystopia, or will it shape a new utopian world?
Emily characterizes AI chatbots as biased “plagiarism machines”, dismissing Sam Altman’s promises that AI will bring “scientific breakthroughs, material abundance and a new chapter in human civilisation…and will soon ‘discover new knowledge’”.
According to Bender, we are being sold a lie: AI will not fulfil those promises, and nor will it kill us all, as others have warned. AI is, despite the hype, pretty bad at most tasks and even the best systems available today lack anything that could be called intelligence, she argues.
An hour and a half later, more than a few otoros deep, George pushes back on the apparent contradiction in Emily’s predictions about AI: “I ask how she squares her twin claims that chatbots are bullshit generators and capable of devouring large portions of the labour market.”
As someone experimenting and building with these technologies, I have to agree with this line of questioning. These generative AI debates feel more like opposing teams vying for influence – battle lines drawn, determinedly arguing past each other. AI, clunky as it is now, will continue getting more and more effective. AI is already reshaping the workforce – just as all earlier waves of technology have. While AI slop is a pervasive and real problem, that won’t be the main event in a few years, either. AI has revolutionized language translation. While initial use-cases have focused on media (text and video content), the ecosystem of models and tooling are extensible far beyond this- to system design, control of tactile devices, autonomous vehicles, robotics and visual analysis –just to name a few. It will take longer than we think to develop the potential of these all things, and that work will surface lots of problems we will need to figure out. As Roy Amara had it, “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”
Comments