TechTakes

https://www.reuters.com/technology/artificial-intelligence/openai-co-founder-sutskevers-new-safety-focused-ai-startup-ssi-raises-1-billion-2024-09-04/

http://web.archive.org/web/20240904174555/https://ssi.inc/

I have nothing witty or insightful to say, but figured this probably deserved a post. I flipped a coin between sneerclub and techtakes.

They aren't interested in anything besides "superintelligence" which strikes me as an optimistic business strategy. If you are "cracked" you can join them:

We are assembling a lean, cracked team of the world’s best engineers and researchers dedicated to focusing on SSI and nothing else.

"Total demand for electricity last year grew by 4.4pc, or 1.3 terawatts (TW), but 80pc of that increase, or 1.1TW, was from data centre growth."

Training data for LLMs = higher energy prices and environmental degradation.

We also want to be clear in our belief that the categorical condemnation of Artificial Intelligence has classist and ableist undertones, and that questions around the use of AI tie to questions around privilege."

• Classism. Not all writers have the financial ability to hire humans to help at certain phases of their writing. For some writers, the decision to use AI is a practical, not an ideological, one. The financial ability to engage a human for feedback and review assumes a level of privilege that not all community members possess.
• Ableism. Not all brains have same abilities and not all writers function at the same level of education or proficiency in the language in which they are writing. Some brains and ability levels require outside help or accommodations to achieve certain goals. The notion that all writers “should“ be able to perform certain functions independently or is a position that we disagree with wholeheartedly. There is a wealth of reasons why individuals can't "see" the issues in their writing without help.
• General Access Issues. All of these considerations exist within a larger system in which writers don't always have equal access to resources along the chain. For example, underrepresented minorities are less likely to be offered traditional publishing contracts, which places some, by default, into the indie author space, which inequitably creates upfront cost burdens that authors who do not suffer from systemic discrimination may have to incur.

Presented without comment.

Got the pointer to this from Allison Parrish who says it better than I could:

it's a very compelling paper, with a super clever methodology, and (i'm paraphrasing/extrapolating) shows that "alignment" strategies like RLHF only work to ensure that it never seems like a white person is saying something overtly racist, rather than addressing the actual prejudice baked into the model.

School student tells AI to put 20 other students’ faces on nude pictures, shares them in chat; it takes months for anyone including the school administrators to act because of some extremely, uh, dubious loophole.

If someone does that in photoshop, it’s a crime; if they do it in AI pretending to be photoshop, it’s somehow not. Gotta love this legal system’s focus on minor technicalities rather than the harm done.