The degree of future acceleration in AI right now is bonkers One day in AI feels like a week in “normal business time” These relativistic effects mean that some companies will accomplish what would normally take 3 yrs in 1 yr And unfortunately, many others will be left behind
@alexandr_wang
-
RLHF: The UI Design Layer for LLM Interaction and Behavior
By
–
RLHF is the "UI/interface design" of LLMs & genAI. It is how we will _design_ the interactions between users & the model, and program it to mirror the desired behavior—helpful, playful, creative, factual Like design, it will be a fundamental way to differentiate in the next era
-
AI Feature vs Product: Retention and Integration as Key Metrics
By
–
toughest questions right now in AI are – is it a feature or a product?
– is it a fad/novelty or a repeat use case? hint: if your retention sucks, it might just be a fad or novelty hint: if an incumbent can easily integrate it, & the best most is “model quality”, it’s a feature -
College Purpose Beyond Academics: Relationships and Community
By
–
the #1 reason to go to college is to find a significant other before the Real World corrupts you and all your suitors the #2 reason is to make friends with whom you can share your whole life arc then it’s a distant 3rd to… great parties?
-
New Mexico Emerging as Next Major Tech Hub
By
–
When I settle down and raise kids, I’ll probably move back to New Mexico. My dream is to be the @rabois of New Mexico. You heard it here first—NM will be the next big tech hub.
-
Great Startups Begin With Hardworking Insecure Nerds
By
–
every great startup I know started as a nucleus of hardworking, insecure nerds
-
CAPEX vs OPEX: AI Infrastructure and Data Spending Economics
By
–
maybe if you compare compute CAPEX vs data OPEX, but if you amortize CAPEX then for major players they are similar orders of magnitude data spend will stay high because while # of hrs goes down, specialization will increase $/hr (and model value) & prod models need validation
-
Scale AI Discusses RLHF Strategy with OpenAI
By
–
btw if you want to chat about your RLHF strategy with us @scale_AI — reach out! DMs are open source on OpenAI:
-
Companies Spending Billions on RLHF Training for LLMs
By
–
we're starting to see top companies spend the same amount on RLHF and compute in training ChatGPT-like LLMs for example, OpenAI hired >1000 devs to RLHF their code models crazy—but soon companies will start spending $ hundreds of Ms or $ billions on RLHF, just as w/compute