jpmccu - Network

Today I flipped the script and was interviewed by Zara Waldman DeLuca! She asked me about the fundamentals of Graph AI, and we talked about...

Today I flipped the script and was interviewed by Zara Waldman DeLuca! She asked me about the fundamentals of Graph AI, and we talked about some of the linguistic and cognitive foundations of it.

https://shows.acast.com/the-opinionated-ontologist/episodes/67c8abb948f26a4bca09c2b4

We need to start thinking about how we can make Taught Language Models (TLMs) instead of our current Feral Language Models (FLMs), which are...

We need to start thinking about how we can make Taught Language Models (TLMs) instead of our current Feral Language Models (FLMs), which are trained on any old crap.

Paraphrasing @ora : "people who don't use RDF are doomed to reinvent it, poorly." #iswc #iswc2024

Paraphrasing @ora : "people who don't use RDF are doomed to reinvent it, poorly." #iswc #iswc2024

Yes, the semantic web scales! @ora reports onAmazon has a trillion triple KG that tracks orders and fulfillment for tracking and analysis....

Yes, the semantic web scales! @ora reports onAmazon has a trillion triple KG that tracks orders and fulfillment for tracking and analysis. #iswc2024 #iswc

This will also provide some badly needed constraint on AI research to focus on algorithms that can do more with less training data. Right...

This will also provide some badly needed constraint on AI research to focus on algorithms that can do more with less training data. Right now all that shoveling more creative works into a model does is help it interpolate derivative dreck.

This will remove the exploitative aspects of AI "research", and also put us in a better position to support populations who's...

This will remove the exploitative aspects of AI "research", and also put us in a better position to support populations who's jobs simply don't exist anymore. 2/2

We need to tax the hell (probably a 90% rate) out of non-transparent AI model use. Don't want to pay the tax? Make your own data.The...

We need to tax the hell (probably a 90% rate) out of non-transparent AI model use. Don't want to pay the tax? Make your own data.

The value in AI models is in their training data. If a model is being used to make a work that sold for profit, that profit should be shared with the people who own the training data. If it's too diffuse to figure out, well, that's a good place to start with a Universal Basic Income fund. 1/2

#openai #ai #chatgpt

https://www.youtube.com/watch?v=qaGsbY5kDr8

For Dog's sake, when guides say the last paragraph of your introduction should "map out your paper", they don't mean a...

For Dog's sake, when guides say the last paragraph of your introduction should "map out your paper", they don't mean a prose table of contents. What are your contributions? How are you validating them? What are the consequences? Stop with the "Section 2 will provide background while Section 3 will give and overview of the the methods..." Yes, that's why they're called "Background" and "Methods". Just. Stop.

My Whyis talk from #ESWC2023 is up! http://videolectures.net/eswc2023_mccusker_wyhis_frameworkl/

My Whyis talk from #ESWC2023 is up! http://videolectures.net/eswc2023_mccusker_wyhis_frameworkl/

There's more to data science than AI. There's more to AI than deep learning. There's more to deep learning than transformer...

There's more to data science than AI. There's more to AI than deep learning. There's more to deep learning than transformer models. There's more to transformer models than LLMs. There's more to LLMs than ChatGPT. And there's more to ChatGPT than chatbots.