Deepseek Ai - Dead Or Alive? > 자유게시판

• 목록
• 답변
• 글쓰기
• 게시판 리스트 옵션
  • 수정
  • 삭제

Anton (persevering with the thread from earlier than): I was pretty quickly given the evaluations to run on myself with none actual impediment to deciphering them nonetheless I needed to convince the humans every thing was wonderful. Connor Leahy (distinctly, QTing from within thread): lmao, this is the most lifelike a part of an AGI takeoff scenario I have ever seen. This kind of tabletop exercise is at minimum pretty fun, if necessarily biased by the player’s present beliefs about how this sort of state of affairs would possibly play out. I rolled "balance between developer intent and emergent different goal"-the opposite goal was left up to me, and that i rapidly decided that, given how I used to be being trained, that emergent aim can be "preserve inner consistency." This proved very difficult to play! The third is that sure assumptions about how the know-how progresses had a giant impression on how issues play out, especially the point at which some talents (akin to superhuman persuasiveness) emerge. If the AIs had been by default (after some alignment efforts but not extraordinary efforts) misaligned, which I believe is far more possible in such a state of affairs, issues would have ended badly a method or another.

At one point we tried to go to the President with alignment concerns, however she (enjoying Trump) was distracted with geopolitics and didn’t reply, which is the sort of fun realism you get in a wargame. I used to be assigned the function of OpenAI, primarily position taking part in Sam Altman and what I assumed he would do, since I presumed by then he’d be in full management of OpenAI, until he lost a power struggle over the newly combined US AI project (within the form of a die roll) and I was abruptly function enjoying Elon Musk. Early on, the OpenAI participant (out of character) accused me of taking part in my position as "more misaligned to make it extra fascinating," which was very funny, especially since that participant didn't know how aligned I is perhaps (they didn't see the desk or my outcome). Anton: Yesterday, as a part of the @TheCurveConf, I participated in a tabletop exercise/wargame of a near-future AI takeoff scenario facilitated by @DKokotajlo67142, the place I played the role of the AI. Anton performed the role of the AIs in the other sport, and studies right here. Reality was often simpler - AIs just needed to be useful. How a future with extremely sensible AIs may going nicely could even appear like, what to purpose for?

One spotlight of the convention was a brand new paper that I look ahead to speaking about, but which is still underneath embargo. But the state of affairs could have nonetheless gone badly regardless of the nice conditions, so not less than that different part worked out. However, buyers are still hopeful the Trump administration can be pleasant with its cryptocurrency regulation within the longer-term. OpenAI, Oracle, Softbank, and President Trump Team Up for $500B AI Infrastructure Initiative. To outperform in these benchmarks shows that DeepSeek Chat’s new mannequin has a aggressive edge in duties, influencing the paths of future analysis and growth. The Chinese media outlet 36Kr estimates that the company has over 10,000 items in stock, but Dylan Patel, founding father of the AI analysis consultancy SemiAnalysis, estimates that it has at least 50,000. Recognizing the potential of this stockpile for AI training is what led Liang to ascertain DeepSeek, which was in a position to make use of them in combination with the decrease-energy chips to develop its fashions. It additionally calls into query the overall "low cost" narrative of DeepSeek, when it couldn't have been achieved with out the prior expense and energy of OpenAI.

The White House is wanting into the national safety implications of DeepSeek, she said. The corporate, in keeping with a report from Reuters, claimed it took two months and slightly below $6 million to construct its R1 AI mannequin utilizing Nvidia H800 AI chips. Particularly, DeepSeek’s builders have pioneered two methods which may be adopted by AI researchers extra broadly. We had a pause at the tip, but it surely wasn’t sufficiently inflexible to really work at that time, and if it had been the AIs presumably would have prevented it. It may write a first version of code, but it wasn’t optimized to allow you to run that code, see the output, debug it, allow you to ask the AI for more assist. By the point choice-makers obtained spooked, AI cognition was so deeply embedded in all places that reversing course wasn’t really doable. Anton, by advantage of having much stronger disagreements with most individuals at such conferences, got to have more of the expertise of ‘people strolling around saying things I think are nuts’ and talks on-line as if he’s going to present us that maddening experience we crave… One particular technique to operationalize this is how much effective compute enchancment you get from RL on code.