Them: "LLMs passing exams shows they are capable of deductive reasoning"
Me: "LLMs passing exams prove that exams do not test for deductive reasoning"
Jon Ayre
@JonHAyre
- 38 Followers
- 17 Following
- 69 Posts
Strategist with a fascination for human dynamics & a talent for tech. Father & Husband. Pessimist & optimist: It WILL go wrong & it CAN be fixed. He/him
James Arthur CattellI'm reading, "Is living in the present the best strategy?" https://jonayre.uk/blog/2022/10/04/is-living-in-the-present-the-best-strategy/ by @JonHAyre from October 2022.
If LLMs prove anything it's that many people are easily fooled into seeing intelligence in eloquently phrased vacuousness.
Show me an AI that can get anywhere near the intelligent action of a cat & maybe I'll listen to your claims of "emerging AGI"
https://jonayre.uk/blog/2023/04/20/the-problem-with-large-language-models/
The problem with Large Language Models - Jon Ayre
Excitement and fear are rife as a result of the recent surge in generative AI. There is talk of LLMs replacing search as an information source, but this ignores one important point. Words, in themselves, convey no meaning. In this respect LLMs are a dead duck.
Ed EverettThe armed Met police that are too worried about illegally shooting unarmed black men to do their jobs can fuck right off.
Their 'strike' is the manifestation of the Met's institutional racism. Sack every one that hands in their weapon, they are labelling themselves as incompetent. #London is safer without them
A point of order if you're claiming the brain isn't a computer.
"It doesn't store memories and process inputs" is nonsense.
"It stores memories and process inputs in a different way to our current CPU based computers" is correct.
The brain is a computer - it's just a fundamentally different type.
Trunk-Based ThierryThis is a marvellous way of looking at a delivery pipeline by Neha Datt and Marcel Britsch #lascot
If you can't stand up for others when they need it, don't call yourself a leader, because you're not.
This is the first real test of leadership.
Adi RobertsonIt’s very funny to me that the dominant Twentieth Century conception of AI was a slightly awkward nerd with an inhuman mastery of facts and logic, when what we actually got is smooth-talking bullshit artists who can’t do eighth-grade math.
Baldur BjarnasonI mentioned here the other day that I keep seeing guys (always guys) being super-impressed about ChatGPT and posting 'brilliant' answers that are factually incorrect.
Turns out that's not even the half of it.
"Scientists, please don’t let your chatbots grow up to be co-authors"
https://garymarcus.substack.com/p/scientists-please-dont-let-your-chatbots
Tarah WheelerHow to phish me:
Send me an email newsletter hyping your
innovative
AI-enabled
blockchain-powered
Web3 product
and include an “unsubscribe” link at the bottom
That’ll do it.