Harlan Stewart and Katja Grace*, 22 December, 2022
Information
New Hires and position modifications
In 2022, the AI Impacts crew has grown from two to seven full time employees. Out of greater than 250 candidates, we employed Elizabeth Santos as Operations Lead, Harlan Stewart as Analysis Assistant, and three Analysis Analysts: Zach Stein-Perlman, Aysja Johnson, and (are within the strategy of hiring) Jeffrey Heninger. We’re excited to have all of them, and you’ll be taught extra about them on our about web page.
Rick and Katja have traded some obligations: Rick is now Director of AI Impacts, and Katja is Lead Researcher. This implies Rick is usually accountable for making selections about working the org, although Katja has veto energy. Katja is answerable for doing analysis, in addition to directing and overseeing it.
Summer season Internship Program
We ran an internship program through the summer time. Between Could and September, six interns labored on varied analysis tasks on matters similar to worldwide coordination, explanations of historic human success, case research in danger mitigation, R&D funding in AI, our new survey of Machine Studying researchers, present AI capabilities, applied sciences which are strategically-relevant to AI, and the dimensions of machine studying fashions.
AI Impacts Wiki
We intend to exchange our pages with an AI Impacts Wiki. Our pages have at all times been functionally one thing like a wiki, so hopefully this new format will make it clearer learn how to work together with them (as distinct from our weblog posts), in addition to simpler to navigate for readers and simpler to replace for researchers. The AI Impacts Wiki will launch quickly and could be previewed right here. . We’ll say extra about different minor modifications after we launch it, however AI Impacts’ previous and future public analysis will probably be both detailed on the wiki or findable by means of the wiki. You may tell us what you suppose utilizing our suggestions kind in addition to feedback on this weblog put up.
Analysis
Completed this yr
This yr, our most important new pages and research-heavy weblog posts are:
AI Impacts is largely a set of pages which are meant to get up to date over time, so our analysis shouldn’t essentially present up as new pages, and is usually a bit more durable to measure than in additional normal analysis establishments. On this event, the above pages and posts most likely symbolize most of our completed analysis output this yr.
In progress
Issues persons are engaged on these days:
- Noteworthy capabilities and limitations of state-of-the-art AI (Zach, Harlan)
- A case research of Alexander Fleming’s efforts to warn the world about antibiotic resistance (Harlan)
- A literature overview of notable cognitive skills of ants (Aysja)
- Assessment and evaluation of AI forecasting strategies (Zach)
- Case research of actors deciding to not pursue applied sciences, regardless of obvious incentives to take action. (Jeffrey, Aysja)
- Strategically important slender AI capabilities (Zach)
- The implications of the fermi paradox and anthropics for AI (Zach)
- What proof from computational irreducibility says about the place highly effective AI shouldn’t be capable of strongly outperform people (Jeffrey, Aysja)
- Proof about how uniform the mind’s cortex is (Aysja)
- Minor additions to Discontinuous Progress Investigation, a challenge searching for historic examples of discontinuous progress in technological traits
- Arguments for AI being an existential danger (Katja)
- An paper concerning the survey (Zach)
- Ending up of assorted summer time internship tasks talked about earlier (interns)
- (Most likely another issues)
Funding
Thanks to our current funders! Together with Jaan Tallinn, who simply gave us a $546k grant by means of the Survival and Flourishing Fund, and Open Philanthropy who supported us for 3 months lately with a grant of of $364,893.
We anticipated to obtain a grant from the FTX Future Fund to cowl working the 2022 Professional Survey on Progress in AI, however didn’t obtain the cash on account of FTX’s collapse. If anybody desires the funders’ share of ethical credit score for paying our survey contributors specifically, at a price of round $30k for the entire thing, please get in contact! (The ex-Future Fund crew nonetheless deserves credit score for substantial encouragement in making the survey occur—thanks to them!)
Request for extra funding
We’d like to be funded extra!
We presently have round $662k. That is about 5-8 months of runway relying on our frugality (spending more cash would possibly seem like e.g. an extra researcher, a 2023 internship program, freer budgets for journey and coaching and so forth, setting salaries extra apt to the job market). We’re searching for one other $395k-$895k to cowl 2023, and would additionally ideally like to increase our runway.
If you happen to may be on this, and need to hear extra about why we predict our work is necessary sufficient to fund, Katja’s weblog put up, Why work at AI Impacts? outlines a few of our causes. If you wish to speak to us about why we needs to be funded or hear extra particulars about what we might do with cash, please write to Elizabeth, Rick or Katja at [firstname]@aiimpacts.org.
If you happen to’d wish to donate to AI Impacts, you are able to do so right here. (And we thanks!)
*Harlan did a lot of the work, Katja put it up with out Harlan seeing its remaining state, so is answerable for any errors.