No. of Recommendations: 28
I am just leaving my 25th business school reunion where I got a chance to sit in on some really great talks about the future of things. Generative AI was a dominant theme at this year's reunion and presented in ways that were humorous and optimistic but also worrisome. As one quote had it, like many technical innovations AI is neither good or bad, but its also not neutral. A few thoughts from the weekend.
* Outside of those deeply invested in the cutting edge of development, most of the world doesn't appreciate how advanced AI is becoming and how much impact it is going to have in a short period of time. A lot of the reason for the underestimation is not because people can't imagine how capable general AI can be "eventually", its because people have a hard time appreciating the time scale at which the advances are happening. The exponential nature of the development happening now means that going from "not as good" to "as good" to "better than human" capability in an area is happening in the span of months, not years like it used to be just 5 years ago.
* AI systems have gotten massively larger recently and encode a lot of context about the world. The context allows them to provide responses that are very sophisticated and allows far greater capacity to infer from limited data the way humans do. It picks up new areas of skill far faster, and at much higher levels of performance to start than humans. We aren't prepared for what's coming in the very near future.
* AI ecosystem has recently (continues to) shifted from a research collabaration paradigm to a economic returns paradigm. The amount of capital required for the largest AI development efforts means money is coming from a lot of sources that want returns on the capital. This is what is happening with OpenAI. With this change in regime, there are some troubling predictions. The focus has changed to one of figuring out who the competitive winners and losers are going to be, and a push to "be first" to claim economic rewards. When that happens, you tend to see a lot more willingness to cheat, break rules, focus on upside and not the downsides, etc. in an incentive system. There is a lot at stake for the world in how things evolve over the next little while and you have to ask who should be looking after safeguards.
* The capacity for efficiency gains is massive. As a company, its not just that you can do the same with fewer people/resources by using AI. In most cases, you will be able to do more with less. It will be a powerful restructuring of how companies organize in areas like software engineering, customer service, planning and analysis, etc.
* Expect there to be a lot of displacement. There has been a saying in radiology, a field that I am familiar with, that AI won't replace radiologists, but radiologists who know how to use AI will replace radiologists who don't. At this reunion, the professors challenged that conventional wisdom. If you could see how AI is starting to function, its hard to conclude that AI won't be coming after a lot of jobs straight up. When cars starting taking over for horses as primary powered transportation, the population of horses plummetted in a handful of years. They didn't become drivers for cars (OK not a great analogy but it was a funny slide they put up to make the point). The speaker made the point that a lot of people today will say that we are so much better off today because of the industrial revolution, but they forget that history is biased by the victors/beneficiaries of a revolution and we are mostly the descendants of groups that benefited. But for 80 years, the industrial revolution caused enormous misery for a lot of the people who were essentially trapped as cogs in the machinery with no way to escape. There are entire classes of people that didn't make it through the revolution. Something similar is likely to be the case with AI.
* Back to those who know how to use AI vs. those who don't. There are scenarios where the distribution of power in society or amongst countries or amongst companies that are as large or larger than countries will shift dramatically. The use of AI by state actors in elections was discussed and detailed work done by a internet security research institute was highlighted as the only reason we have been able to trace that these things happened. (BTW, that institute's funding was recently gutted by the efforts of a congress person who wanted to investigate them because they highlighted russian efforts to tilt the race toward republicans). But it was fascinating to see how AI allows really rapid, scaled, inexpensive AB testing to find out what works and doesn't on social media to rouse us a group and create an us vs. them narrative. Some of the examples seem non-sensical as humans, but effective if you remove our human biases. Can you run a credible newsservice that serves 50M people in rural India from a Chinese editorial HQ where no one speaks any Indian language? Apparently you can if you can build the AI tools to do all of the gathering, contextualizing, and distribution for you. Can a group in Russia figure out how to get into the inner dynamics of a particular "identity group" like southern christians, or alternatively liberal bay area techies, and figure out how to be really effective at creating viral, influential content tailored for those groups? No problem. Some of the examples were eyeopening for sure.
How these changes coming down the pipe will affect BRK, I don't know. But I've been through the hype cycle with both the dot com era and now AI. This one actually feels far more substantial and based in demonstrated examples of capabilities enabled than the hype around dot com. And no one can deny how much the things that came out of the dot com era (separate from investment returns) have radically changed our world.
