Just out - ongoing predictions for AI. Are they in sight after ~70 years of trying?
Beth Carey
co-founder of machine intelligence company Pat Inc - creators of machine readable meaning for the world's languages
Beth Carey boosted:
Didja read about the study that claimed ChatGPT produced poetry “indistinguishable” from Shakespeare?
Critique by NYU’s Ernest Davis, on the unbearable banality of ChatGPT’s poetry: https://cs.nyu.edu/~davise/papers/GPT-Poetry.pdf
Beth Carey boosted:
Tonight (Monday) in Seattle Town Hall, Gary Marcus and Ted Chiang talk AI and Cognitive Science:
https://townhallseattle.org/event/gary-f-marcus-with-ted-chiang/
I'm enjoying the thumbnail previews that @Mastodon create unlike some other platforms
Where do you think the next generation of AI tools will come from?
"the 2010s were the age of scaling,"
https://arstechnica.com/ai/2024/11/what-if-ai-doesnt-just-keep-getting-better-forever/
@garymarcus I asked the question if others, like Steven Marlow, found 'that other platform' lonelier now for AI writers since Gary left ......
@garymarcus Apparently not one LLM company responded to UK's guidelines (soft 'regulation') set forth this year to provide the training data and weights.
Beth Carey boosted:
On diminishing returns in LLMs and why OpenAI might become the WeWork of AI:
Beth Carey boosted:
Discussing the mysterious absence of GPT-5 with @LizClaman @FoxBusiness around 320pm ET.
Tired of the tech giants' narratives based on their agendas?
For the trustworthy language interface of the future, cognitive scientist John Ball talks about his viable alternative in recently published book 'How to solve AI with Our Brain',
Beth Carey boosted:
Check out this really excellent piece from @abebab and @rajiinio if you haven't: https://www.wired.com/story/large-language-models-critique/
Lots of great things to quote in here, but this definitely sums it up:
"If these powerful companies cannot release systems that meet the expectations of those most likely to be harmed by them, then their products are not ready to serve these communities and do not deserve widespread release."
It also happens that enterprise sales and service only do goal oriented conversations requiring accuracy and reliability. Not to mention an even higher bar, mission critical industries like medical, finance and legal.
RT @ylecun@twitter.com
@Hello_World@twitter.com The fact that LLMs "make stuff up" is not a bug if you are using them to write fiction.
It is if you are using them to search for information, answer questions, or write non-fiction.
“The current paradigm around brain function is that they are prediction machines — a concept that explains nothing about how they work. Rather than prediction, pattern-matching can be learned by experience while giving the appearance of a…” — @jbthinking@twitter.com https://link.medium.com/l0EffDIJYwb
“Patom Theory (PT) is my theoretical neuroscience model that explains how brains pattern match sets and lists of elements, in a hierarchical, bidirectional manner. It explains brain function and considers the observations in the Hawkins’…” — @jbthinking@twitter.com https://medium.com/pat-inc/2023-the-year-of-understanding-theory-f55692f9add2?source=social.tw
“Linguistics provides a process for humans to eliminate ambiguity. Our brains learn that the way in which objects are referred to is restricted by their meaning” — @jbthinking@twitter.com https://link.medium.com/njOCdKsVXwb
“This makes the NLU industry ‘come of age’: when you can use a system without configuration for conversation, NLU will be the default way to work.” — @jbthinking@twitter.com https://link.medium.com/IgxUCYD6Rwb #NLU #NLP #AI
' Implicit in today’s platforms is the requirement for ongoing training, on top of the initial training. ..... the cost of training is expensive. 1 of the tech giants we met mentioned that they can spend $m's to create a corpus to use in machine learning! https://buff.ly/3kEiJEj
Client Info
Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst