The new CMIP7 emissions scenarios are out for the IPCC's seventh assessment report (AR7). See the CarbonBrief explainer for a full breakdown.
Notable changes:
The highest 8.5 scenario was dropped and the now "high" scenario has much less warming
Much of this is because of climate policy and low cost tech; high coal use for electricity is no longer plausible given the low cost of renewables and batteries
No baseline scenario; instead they show less/current/more climate policy, with a medium scenario with current policy at around 2.9°C of warming
Scenarios still depend on a lot of carbon-dioxide removal that hasn't scaled yet
The lowest (1.5°C) and highest (4°C+) scenarios are less feasible. That's mixed news - less chance of the worst case scenarios, but also means we are more likely to still have disruptive ~2.5-3°C warming.
Headline finding: I audited 17 AI Safety Talent programmes. Zero of 17 have published any comparison group, rejected-applicant follow-up, matched control or randomisation. Not one. Every programme that mentions a counterfactual does it by asking participants to self-report.
Background
At least $70 million...
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...
A year ago I posted that YCombinator (YC) companies didn’t seem to be growing faster since the release of ChatGPT in 2022. I reran that experiment and found that 2023+ YC companies are arguably growing a bit faster than pre-2023 companies (including AfterQuery, ...
