AI - AI Dynamics

RLHF preference for direct answers over clarifying questions

By

–

23 January 2023 10h53

I suspect this is because of RLHF — if the goal is to choose responses a human would rank highly against other options, “Could you please clarify X?” or “Am I understanding right that you want Y?” are not usually major contenders for top place.

→ View original post on X — @saboo_shubham_

Right Partner Makes Founder Success 100x Better

By

@saboo_shubham_

–

23 January 2023 10h45

Dating the right person makes you 100x founder

→ View original post on X — @sallyeaves

New Digital Twins Development Resource Emphasizes Inclusion

By

@sallyeaves

–

23 January 2023 10h34

Always appreciated @MikeNashTech & apologies I missed this reply – have just added a new resource re #DigitalTwins development & #inclusion – some excellent assets here! Many thanks S

Chat-tuned LLMs hover between conversation and instruction prompting

By

–

23 January 2023 10h30

Today’s “chat” is somewhere between conversation and instruction prompting. Chat-tuned LLMs never ask you for more information, or quickly confirm their interpretation of a complex task before starting. The lineage shows.

→ View original post on X — @officialindiaai

AI-Guided Methods Discover New Nanostructures at Brookhaven Lab

By

@officialindiaai

–

23 January 2023 10h17

Did you know that AI-guided autonomous methods can discover new materials like nanostructures? Tap here to know the startling discoveries made by scientists at the US Department of Energy's Brookhaven National Laboratory: https://
bit.ly/3XS5A8O

Image models prompting: associative bag of labels

By

–

23 January 2023 9h53

I don’t know much about image models, but prompting them seems very different. It’s more of an associative bag of labels than a text.

→ View original post on X — @officialindiaai

Human-Computer Interaction: Usability and User Experience in Computing

By

@officialindiaai

–

23 January 2023 9h30

Human-Computer Interaction is the latest buzzword in IT. Tap here to explore this branch of computer science that examines how people interact with computers and understand its usability: https://
bit.ly/3XwdWDk

Just Ask is Baseline, Not Competition to Prompt Engineering

By

–

23 January 2023 9h25

I don’t think “Just ask” is competition to prompt engineering — it’s the baseline that it extends. If our oversight stops being useful, it implies we can no longer imagine a task too hard for “Just ask” to do reliably. This seems unlikely pre-AGI.

Prompt engineering valuable until AGI; ‘just ask’ will lag behind

By

–

23 January 2023 9h25

People ask me about the ephemerality of LLM prompt engineering. I think it will be valuable for the foreseeable future, i.e. until AGI. “Just ask” will improve. Today’s PE with be obsolete, perhaps quickly. But “Just ask” will always lag behind the frontier.