One of the biggest areas of AI hype right now is the notion that it will hyperaccelerate scientific progress. I understand why people think this — AI is already accelerating scientific production. But the big problem is that production seems to have gotten decoupled from any useful measure of progress a long time ago. There is a whole science-of-science literature hand wringing about the fact that scientific production has increased ~exponentially but progress has not accelerated and has even slowed down. Producing papers, for the most part, is a game researchers must play for status and career progress. It's value is relative. It's like thinking that AI is going to help traders make a lot more money. If everyone has access to the same capabilities, there is no alpha. In every scientific field I'm familiar with, the amount of published stuff exceeds the community's collective bandwidth to absorb and build upon ideas by a factor of 100x or more. Inevitably, the vast majority of what's published makes zero impact. Yet we pretend that publication itself has some value. It doesn't. I do think AI will help, but ironically, it's by stretching an already creaking system to its breaking point and forcing it to reinvent itself. For example, AI is getting to the point where it is able to produce papers that aren't easily distinguishable from human-authored ones. Dropping the cost of production to zero will force us to stop attaching value to publication, and look for ways to identify actual intellectual value. There's also the fact that in many/most fields, there's a limit to AI's impact on production as well. It may be able to do the cognitive parts but the bottleneck may be experiments on humans or some other form of real-world interaction that's hard to automate.
I think the biggest potential for LLMs in the research space is helping to read and summarize the vast flow of papers. I am curious if an LLM can answer the question “what is new” in a paper, since this requires an understanding of the relevant papers in a field.
When the browser came out in 1990, it was clear that printed scientific journals would become a thing of the past. Still, it was a slow process before the printed page became irrelevant. I remember vividly how exciting it was, and also how stubborn people were to accept change. As AI progresses, what we know as 'papers' today will change, and science will be written instead for the AI to consume. AI will soon be listening to and even watching lectures, doing homework problems, and learning what is new globally at a lightning pace. Researchers will stop reading (as if they read a lot now) and instead consult the AI to learn what is new. The nature of authorship itself will change, as the AI copilots will be used to help discover and formulate new ideas. And, eventually, as technology improves, AI will conduct simulations and robots will help perform lab experiments. Imagine telling a PhD student today to go to the library to photocopy an article. Those days have long gone. The researchers of tomorrow will likewise live in a very different world. And we are just getting the glimpses of this now.
There is a deep insight in this post, "the amount of published stuff exceeds the community's collective bandwidth to absorb and build upon ideas by a factor of 100x or more" There are real limits to how much we as humans can absorb. This applies in terms of knowledge but it also applies to the amount of change we can accept. We have a budget of attention, once that budget is exceeded extra thing's aren't relevant.
Unfortunately I agree. GenAI slop is now everywhere, and the society at large is worse off because of that. That is why overly optimistic coverage of these tools is so harmful - it incites too many people to deploy them without thinking about limitations and consequences.
Why relating it to trading? AI for science is an alpha for humanity. To your point, if everybody uses it, then everybody goes faster, which means no individual gets an edge but we all benefit from it.
Arvind Narayanan its truly mystifying why conversation about AI seems focused on advances on generative AI. Do look at the work thats being done by George Karniadakis , Anima Anandkumar , DoE labs in scientific machine learning. Those advances are not trivial. AI space can in general do with less hype, but surely don't throw the baby out with the bath water
We're drowning in information, but starving for wisdom. It's not simply that AI has & will accelerate the production of scientific papers; it's that it has & will exacerbate a pre-existing crisis of meaning. The exponential increase in scientific output hasn't led to proportional progress because the system of value is broken. We've confused activity with achievement.. This applies to all knowledge creation avenues. We are moving from an era of information scarcity to one of meaning scarcity.
AI’s potential to accelerate scientific progress is real but the system needs a reset. ✅ Exponential production ≠ progress. The gap between publishing papers and creating meaningful impact is widening. ✅ AI could break the system. By flooding it with content, it might force us to rethink how we measure intellectual value. ✅ The real bottlenecks? Experimental validation and real-world applications not just the cognitive tasks AI excels at. 💡 AI is only as impactful as the system it works within. How do we reinvent this system for true progress?
In recent times the length of scientific articles also have grown exponentially, at least in AI. Look at the technical report of gpt4 (100 pages !) which has hardly anything “technical” - it is purely marketing material ! The same is true for other papers also. Most performance metrics look highly subjective & most claims cannot be verified. It would be fair if benchmarking of models is done by independent groups and not by the group producing models.
Executive Chairman and CEO
40mThis is an insightful take on the relationship between AI, scientific production, and progress. While AI undoubtedly accelerates scientific output, it’s crucial to recognize that mere quantity doesn’t equate to true innovation. We must shift the focus from publication counts to actual intellectual impact. It’s interesting to think about how AI could eventually force a rethinking of what we value in science. Perhaps it’s time for a reinvention in how we approach and measure scientific progress. 💡 On a related note, as a digital marketing agency, we work closely with cutting-edge technology to help businesses not only grow but also focus on real value. Interested in how AI can drive smarter marketing and real results? Let’s connect! #AI #Innovation #Science #DigitalMarketing