A guide for the perplexed on AI

{"text":[[{"start":5.15,"text":"How should someone who knows next to nothing about artificial intelligence think about its implications for humanity? While it is cheeky to use the formulation of the Jewish sage Maimonides when tackling the relationship between revelation and philosophy, it is not absurd. After all, even the greatest sage cannot fully understand divinity. So, the fact that I do not understand the implications of AI should not prevent me from struggling to do so. Maybe my struggles will also help others."}],[{"start":35.6,"text":"So, here goes. The question I wish to consider is: “AI boon, bane or bubble?” Moreover, once we have considered the answers, are there any choices humanity can realistically make to ensure it is far more of the first than the second, or are we doomed to be dragged behind the AI chariot, willy-nilly, wherever it goes?"}],[{"start":null,"text":"

"}],[{"start":55.55,"text":"The answer to the question about whether it is merely a bubble will help to set the ground for answering these as well. So, what might it mean for AI to be a bubble? There are two possibilities."}],[{"start":67.45,"text":"One is that something important is indeed happening. But markets cannot estimate the returns and are being driven into a speculative frenzy. This, in turn, is fuelling an unsustainable (and to some degree unprofitable) surge in investment. Sooner or later, this bubble will burst, stock prices will collapse, many old and new businesses will go bankrupt and the investment will subside. But we will be left, as we were after, say, the railway booms of the 19th century and the dotcom bubble of the 1990s, with useful infrastructure: tracks, in the case of the former, and fibre optic cable, in the case of the latter. Such bubbles can transform the world. "}],[{"start":null,"text":""}],[{"start":107,"text":"The other possibility is that AI is nonsense. The Mississippi and South Sea bubbles of the early 18th century in France and England come to mind: they burst, ruined some, and changed little."}],[{"start":118.05,"text":"So, is what is happening to AI a bubble at all and if so of which kind? The consensus, which I (tentatively) share, is that AI is real. Whether we are on the threshold of artificial general intelligence, as Demis Hassabis of Google DeepMind argues, I have no idea. But today’s models seem impressive, especially in their “agentic” role."}],[{"start":141.95,"text":"Moreover, as VC investor Rubén Domínguez Ibar notes in his newsletter The AI Corner, some providers, notably Anthropic, are generating huge rises in revenue, which helps explain the huge valuations expected in initial public offerings. So, the market is indeed bubbly, but it appears based on reality."}],[{"start":null,"text":""}],[{"start":161.7,"text":"Some have made comparisons between the spectacular surge in the shares of Nvidia with the not dissimilar performance of Cisco in the dotcom bubble. But they note a difference: Nvidia’s earnings have soared, while Cisco’s net income merely doubled in the two years to July 2000. Moreover, analysis from the Peterson Institute for International Economics suggests AI is already generating a huge hidden jump in US real GDP."}],[{"start":190.35,"text":"It is impossible to say the markets are “right” in judging the profitability of the current and prospective AI champions. It is also possible that the earnings we are seeing in Nvidia (or even the revenue in Anthropic) will not be sustained when the investment surge and AI-hype subside. But AI is not just a bubble, it has reality."}],[{"start":212.7,"text":"This then brings us to “the boon or bane” question. I had a discussion with ChatGPT (what else?) about this. After some to-ing and fro-ing between us, we ended up listing the top boons and banes as follows."}],[{"start":null,"text":""}],[{"start":225.2,"text":"First, the boons: better healthcare; an acceleration of science; far higher productivity; education at scale; faster progress on climate and clean energy; improved accessibility and inclusion (speech-to-text and automatic translation); improved public services; better transportation and safer work environments; improvements in creativity and cultural expression; and improved access to human knowledge worldwide."}],[{"start":251.64999999999998,"text":"Then, the banes: loss of human control and accountability; lethal new weapons (notably pathogens, but other weaponry too, some in the hands of terrorists); mass unemployment; vast concentrations of power in the hands of monsters; mass surveillance and authoritarian control; even more misinformation and manipulation; huge threats to cyber security; entrenchment of bias and discrimination masked as “objectivity”; erosion of human agency and skill; and the environmental costs of vastly resource-intensive systems."}],[{"start":null,"text":""}],[{"start":284.5,"text":"What is one to make of these (entirely plausible) lists? My first conclusion is that AI is not just any “general purpose technology”. If it develops in the ways that seem likely, it might change almost everything: AI is indeed existential. If humanity had any collective sense (which it does not) and had the capacity to stop itself (which it does not), it would, I think, pause the whole thing. "}],[{"start":310.1,"text":"Second is that AI has inevitably unleashed a competition among businesses and governments. My assumption is that anything such a technology can do will then be done. So, we are going to see competitive races for both the boons and the banes."}],[{"start":null,"text":""}],[{"start":326.75,"text":"My third is that the relatively successful controls on nuclear proliferation and the availability of new drugs cannot be precedents. That is because AI will not be the property solely of states, as nuclear weapons have (so far) been, and it is not a single class of products, like drugs. It is polyvalent."}],[{"start":344.7,"text":"ChatGPT suggested humanity should stop treating technological capability as the same thing as progress, which should mean flourishing in conditions of “safety, liberty and legitimacy”. Pope Leo agrees: build “for the common good”, he says in his encyclical. But is this possible and if so, how? I will consider that next week."}],[{"start":371.1,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1780484448_2234.mp3"}