I don't feel entirely comfortable writing a review of Mustafa Suleyman's book about the future of AI and other disruptive new technologies because I mostly just skimmed it. In my defense, it's one of those books that was essentially written to skim, the sort of long-winded business book that rehashes historical anecdotes you've heard countless times and speaks in vague but eminently digestible abstractions. In a functioning market, all of these books would be reduced to the longish magazine article that comprises their core. But how do you make a living on magazine articles? So I will forge ahead with some comments despite my reservations since I did go through the whole book and I think I managed to extract the essential gist.
Sulyeman, as the founder of DeepMind, is certainly qualified to opine on the technology of AI. This is not a guy you can dismiss as not understanding how the stuff actually works. As you might expect from such a guy in our current moment, he is wildly optimistic about the technical possibilities. He thinks AI will facilitate revolutionary new developments in many areas, particularly in biology. To his credit, his technical optimism is based in concrete thinking about extensions of what these technologies can already do, not some pie in the sky thinking about what happens when a magical AGI appears on the scene.
One thought he had in particular has stuck with me as a summary of this optimism -- his off-handed certainty that very soon, you will simply be able to prompt an AI to start a business for you selling things on Amazon that turns $100,000 in seed capital into $1,000,000 in profit. This agent would then go out there and do some consumer research, incorporate itself, create a product design, arrange for some manufacturing, and list this new must-have item on Amazon. Shocking as the example sounds at first, in a way, this optimism strikes me as entirely plausible. After all, it would be great for Amazon if this existed, and we all know who is really running the show these days. In fact, the description is pretty close to what already passes for "innovation" in entrepreneurial circles circa 2025, with the possible caveat that the things sold should really be software, rather than a physical product, because it scales better and has higher margins. In other words, we've already created all the conditions that would require an agent like this to thrive. All that remains is to automate a well known process.
What makes Suleyman's book more interesting than other breathless celebrations of the potential of AI is that he spends quite a bit of time taking the next step and examining the world such tools would create. While you will have to consult other sources for a description of what might happen in the particular example cited, Suleyman discusses many ways in which AI tools could be used to undermine state authority, exacerbate misinformation, and entrench inequality. So the book is mostly meant as a warning about how we need to adapt to the current moment and pro-actively consider some of these consequences. Unfortunately, most of the solutions he suggests would require a highly functioning government or very broad minded technologists such as himself, and even then they would be difficult to implement. And since he rightly feels that both of these are long shots, his technical optimism is the cause of his profound social pessimism. Fundamentally, Suleyman sees these technologies changing society in highly disruptive and unpredictable ways, and he feels, based on our historical experience with technology, that we simply won't be able to control the impacts. To many parts of our social structure will see every incentive to barrel ahead regardless of the consequences. Businesses will see better profits. Governments will see better control. Malicious actors will see better weapons. And the rest of us will see what we have by now gotten pretty used -- enshittification wrapped up as shiny progress porn, an endless scroll of exploitation.
So we should give the book credit for raising some inconvenient questions about technologies the author has had a considerable hand in creating. And perhaps we should not go too hard on our brave entrepreneur turned Microsoft exec just because he overlooks the only plausible solution -- a collective reckoning with those aspects of our technology that have already been working against human flourishing, and which look set to use these new tools to further flourish at our expense.
Sulyeman, as the founder of DeepMind, is certainly qualified to opine on the technology of AI. This is not a guy you can dismiss as not understanding how the stuff actually works. As you might expect from such a guy in our current moment, he is wildly optimistic about the technical possibilities. He thinks AI will facilitate revolutionary new developments in many areas, particularly in biology. To his credit, his technical optimism is based in concrete thinking about extensions of what these technologies can already do, not some pie in the sky thinking about what happens when a magical AGI appears on the scene.
One thought he had in particular has stuck with me as a summary of this optimism -- his off-handed certainty that very soon, you will simply be able to prompt an AI to start a business for you selling things on Amazon that turns $100,000 in seed capital into $1,000,000 in profit. This agent would then go out there and do some consumer research, incorporate itself, create a product design, arrange for some manufacturing, and list this new must-have item on Amazon. Shocking as the example sounds at first, in a way, this optimism strikes me as entirely plausible. After all, it would be great for Amazon if this existed, and we all know who is really running the show these days. In fact, the description is pretty close to what already passes for "innovation" in entrepreneurial circles circa 2025, with the possible caveat that the things sold should really be software, rather than a physical product, because it scales better and has higher margins. In other words, we've already created all the conditions that would require an agent like this to thrive. All that remains is to automate a well known process.
What makes Suleyman's book more interesting than other breathless celebrations of the potential of AI is that he spends quite a bit of time taking the next step and examining the world such tools would create. While you will have to consult other sources for a description of what might happen in the particular example cited, Suleyman discusses many ways in which AI tools could be used to undermine state authority, exacerbate misinformation, and entrench inequality. So the book is mostly meant as a warning about how we need to adapt to the current moment and pro-actively consider some of these consequences. Unfortunately, most of the solutions he suggests would require a highly functioning government or very broad minded technologists such as himself, and even then they would be difficult to implement. And since he rightly feels that both of these are long shots, his technical optimism is the cause of his profound social pessimism. Fundamentally, Suleyman sees these technologies changing society in highly disruptive and unpredictable ways, and he feels, based on our historical experience with technology, that we simply won't be able to control the impacts. To many parts of our social structure will see every incentive to barrel ahead regardless of the consequences. Businesses will see better profits. Governments will see better control. Malicious actors will see better weapons. And the rest of us will see what we have by now gotten pretty used -- enshittification wrapped up as shiny progress porn, an endless scroll of exploitation.
So we should give the book credit for raising some inconvenient questions about technologies the author has had a considerable hand in creating. And perhaps we should not go too hard on our brave entrepreneur turned Microsoft exec just because he overlooks the only plausible solution -- a collective reckoning with those aspects of our technology that have already been working against human flourishing, and which look set to use these new tools to further flourish at our expense.
No comments:
Post a Comment