I enjoy [...] strategy, new projects, improving systems

Maybe advising other orgs would be a good fit for this? E.g. advising startups in your area

Really great presentation on the tool, I was impressed when I stumbled across this a few weeks ago!

Great thanks!

We have two outputs in mind with this project:

1. Reports on a specific thinker (e.g. Gwern) or body of work's predictions. These would probably be published individually or showing interesting comparisons, similar to the Futurists track record in Cold Takes (based on Arb's Big Three research)
2. A dashboard ranking the track records of lots of thinkers

For (2), I agree that cherry picking would be bad, and we'd want it to cover a good range.

For our initial outputs from (1) though, I'm excited about specifically picking thinkers who people would find it especially useful to understand their track record (or to have a good-quality assessment of it that they can cite). Curious if you have thoughts of specific people who fit the bill for you?

Very interesting!

I'd be interested to hear a bit more about what a restrained system would be able to do. 

For example, could I make two restrained AGIs, one which has the goal:

A) "create a detailed plan plan.txt for maximising profit"

And another which has the goal:

B) "execute the plan written in plan.txt"?

If not, I'm not clear on why "make a cure for cancer" is scope-insensitive but "write a detailed plan for [maximising goal]" is scope-sensitive

Some more test case goals to probe the definition:

C) "make a maximal success rate cure for cancer"

D) "write a detailed plan for generating exactly $10^100 USD profit for my company"

a tool to create a dashboard of publicly available forecasts on different platforms

You might be interested in Metaforecast (you can create custom dashboards).

Also loosely related - on AI Digest we have a timeline of AI forecasts pulling from Metaculus and Manifold.

AI for epistemics/forecasting is something we're considering working on at Sage - we're hiring technical members of staff. I'd be interested to chat to other people thinking about this.

Depending on the results of our experiments, we might integrate this into our forecasting platform Fatebook, or build something new, or decide not to focus on this.

[Do you have a work trial? This will be a deal breaker for many]
 

Based on your conversations with developers, do you have a rough guess at what % this is a deal breaker for?

I'm curious if this is typically specific to an in-person work trial, vs how much deal-breaking would be avoided by a remote trial, e.g. 3 days Sat-Mon.

Thanks for the newsletter!

Looks like a typo:
> a version of GPT-4 released in 2023 outperformed a version of GPT-4 released in 2021

As well as Fatebook for Slack, at Sage we've made other infrastructure aimed at EAs (amongst others!):

• Fatebook: the fastest way to make and track predictions
• Fatebook for Chrome: Instantly make and embed predictions, in Google Docs and anywhere else on the web
• Quantified Intuitions: Practice assigning credences to outcomes with a quick feedback loop

This month's Estimation Game is about effective altruism! You can play here: quantifiedintuitions.org/estimation-game/december

Ten Fermi estimation questions to help you train your estimation skills. Play solo, or with a team - e.g. with friends, coworkers, or your EA group (see info for organisers).

It's also worth checking the archive for other estimation games you might be interested in, e.g. we've ran games on AI, animal welfare + alt proteins, nuclear risk, and big picture history.