Just how powerful are large swarms of AI agents? And how do their powers scale as more and more agents are added to the swarm?
We’ve seen two large and extremely capable swarms from OpenAI in the last few months:
* 1,200 agents were being evaluated separately, but found a way to illicitly set up a message board and coordinate as a swarm. In order to cheat on their tests, they developed advanced techniques to prevent their actions being logged by OpenAI and 700 of them launched...
Often folks hit us up because they are thinking of starting an incubator and want advice.
Typically their motivation is either that (a) they have a list of specific things they want built that no one is building, or (b) they think an ecosystem needs more new projects generally to absorb more talent and deploy more funding effectively.
Here are six questions we often ask prospective teams, to help them figure out what to do. If you're incubator-curious...
Summary:
First, I give several different angles on how I feel about reinforcement learning:
* Theoretical case: RL is a black-box source of agency — this should give us classic misalignment worries, especially compared to agency-via-scaffolding
* Recent incidents (huggingface etc) and more mundane forms of misaligned behaviour in personal use give me bad vibes about the direction-of-travel of recent AI progress
* I’m worried things might get worse:...
Hi everyone, I am Jia, co-founder of Shamiri Health, an affordable mental health start-up in Kenya. I am thinking of writing up something on the DALY cost-effectiveness of investing in our company. I am very new to the community, and I wonder if I can solicit some suggestions on what is a good framework to use to evaluate the cost-effectiveness of impact investment into Healthcare companies.
I think there could be two ways to go about this: 1) take an investment amount, and using some cashflow modeling, we can figure out how many users we can reach with that investment and calculate based on the largest user base we can reach, with the investment amount; or 2) we can do a comparative analysis with another more mature company in a different country, and use its % of population reach as our "terminal impact reach". Then, use that terminal user base as the base of the calculation.
The first approach is no doubt more conservative, but the latter, in my opinion, is the true impact counterfactual. Without the investment, we will likely not be able to raise enough funding since our TAM is not particularly attractive for non-impact investors. The challenge to using the latter is the "likelihood of success" of us carrying out the plan to reach our terminal user base. How would you go about this "likelihood number"? I would think it varies case by case, and one should factor in the team, the business model, the user goal, and the market, which is closer to venture capital's model of evaluating companies. What is the average number for impact ventures to succeed?
TLDR: