Trust, reliability, and safety in AI ft. Daniela Amodei of Anthropic and Sonya Huang
Daniela Amodei, co-founder and president of Anthropic, sits with with Sonya Huang at Sequoia Capital's AI Ascent to discuss the launch of Claude 3, solving the business problems of trust and reliability in AI, the importance of transparency in research, and implementing technical safety approaches to make AI more aligned with the values of the human race.
#AI #AIAscent #Sequoia #Startup #Founder #entrepreneur
Пікірлер: 33
please change the thumbnail is Daniela Amodei, not Amoedi
Thank you Sequoia
Bravo Anthropic Team
What a great and useful service, thanks
Amazing talk
Great that you record these given the slim attendance
A rising boat lifts all tides! :-) 🤣
We sure down want the woke AI road 👍
By reading the comments and the Anthropic wiki page it's very clear that someone is paying a campaign to diminish the impact of this company, their products and their members. As someone that has tried various online models as well as locally running open source ones, it's clear that they have a great product with Claude that has a great price, performance, and usability ratio. It's no surprise that they have been raising a bunch of capital from predominant players in the tech industry. It's also great to see a female founder from which some young girls can get inspired to pursue a technical interest. Thank you for sharing this talk.
@ClimateDS
2 ай бұрын
> it's very clear that someone is paying a campaign to diminish... Are you serious? To me, that sounds like something someone would say when trying to diminish their own cognitive dissonance.
I read the thumbnail as Daniela AI model
THEY tailor make their own and its all how you organize their data,and some need to be with their profile ,and for their different task/multifaceted life and personality.Look for ''efficient'' and it will solve the other problems ,create and think silos and branching at the tops in some extremes/work research development etc.''create efficiency'' while solve the other problems its win win win lmao
Gpt4 is like the ex that only could lay there lazily, but got the job done, while claude has a cowboy hat riding like the wind bulleye.
Empty chairs? I would have thought it would be a full house.
@clray123
2 ай бұрын
That lady does not appear very technical or knowledgeable about anything, maybe that's why.
How can hallucination become zero !! It’s a probability machine.. it can never be perfect! The video was more focused on advertising Anthropic rather than talking novel or new things..
This entire video is just one big ad. There's no useful information in it.
@3lbios
2 ай бұрын
yeah. just compare to the Mistral CEO video, wildly different communication approaches
@ClimateDS
2 ай бұрын
Indeed. it's annoying but shouldn't be surprising. As she said countless times, they target Enterprise customers. Mistral on the other hand wants to serve the AI community.
@andybaldman
2 ай бұрын
@@ClimateDS AI as a whole is ultimately primarily set to benefit corporations (i.e., entities that deal in big data), not individuals. Despite all the hot air we're hearing from AI CEO's, AI isn't anywhere near as beneficial to individuals as Google was 25 years ago. Once upon a time when Google was new, everyone young and old was googling stuff. Literally my parents and grandparents. Today, 99% of people have no use for the AI tools that are being rolled out, unless you're a video creator who doesn't want to pay for stock images/video, or if you want to save some time making content. However 99% of people don't make online content. They just consume it. The real power of AI is in helping corporate managers cut costs (and people) through more automation. And there are going to be a lot of people pushed out of the workforce as a result, with no place else to go. Because you can't say 'well, people will just learn new skills for new jobs', when LEARNING itself is what's being automated.
@ClimateDS
2 ай бұрын
@@andybaldman Indeed, the end user will be mainly corporations but I get the impression that the difference here is that claude is targeted directly to the devs within those corporations, while Mistral models targeted to devs in SaaS/startups, which sell LLM-based products to corporations.
@andybaldman
2 ай бұрын
@@ClimateDS Agreed, well put.
Not Mentioned: Why you can't beat a year old model with your latest model. 🤔
@aloysius_music
2 ай бұрын
Opus outperforms GPT-4 on most evals & lmsys arena though?
@Alice8000
2 ай бұрын
@@aloysius_music show me
Ummmm Didn't she just get arrested with SBF for doing the crypto scam???
I find this entire video a whole lot of "interesting"
and you can only tackle so many problems from ''one front','multi-layers and from different angles and planning is need ,broke down problem.HIRE A ''CIA analyst''' WHO CAN USE some intuition/commonsense Im sure and they take of 90% of your problem its their job and control the probability's or ''mitigation'' lmao
Sequoia is in big trouble. Very superficial discussion