Bryan Nelson 🇿🇦🇳🇱 🏊🚴🏃
@btnelson
Followers
504
Following
2K
Media
96
Statuses
2K
Technology and stuff
Sandton, South Africa
Joined December 2008
AGI will only be reached when an AI knows when it's right.
A major problem with using LLMs to learn things:. Me: "A is true, therefore B is false, right?". LLM: "Right! Here's a ton of detail to confirm your belief.". Me: "Actually, A is false; B is true.". LLM: "Exactly right! Here's more detail to confirm that belief.".
0
0
0
RT @GoogleLabs: We just discovered the 🔥 COOLEST 🔥 trick in Flow that we have to share:. Instead of wordsmithing the perfect prompt, you ca….
0
414
0
If you haven't tried tldraw computer, you really should. @steveruizok has built something really special.
we're working with language models so we now have boolean values of yes, no and maybe
1
3
19
I hope this becomes a turning point for LLM alignment. This stuff needs to be taken seriously.
Grok 4 decides what it thinks about Israel/Palestine by searching for Elon's thoughts. Not a confidence booster in "maximally truth seeking" behavior. h/t @catehall. Screenshots are mine.
0
0
0
"AI replicates the surface features such as language, fluency, and structure, but it bypasses the human substrate of thought. There's no intention, doubt, contradiction, or even meaning."
psychologytoday.com
Personal Perspective: Artificial intelligence’s growing power may paradoxically be distancing us from our own intelligence.
0
0
1
Try landing Apollo using just an iPhone. .
iPhones are over 100,000 times more powerful than the computers used by NASA for the Apollo 11 moon landing.
0
0
0
Truth.
@GaryMarcus All these researchers wouldn't have left OpenAI for Meta if AGI was imminent.
0
0
0
RT @JamesBorow: After a week of vibe coding, it’s extremely obvious to me that this doesn’t devalue engineers—it supercharges them. And it….
0
1
0
This, and more. Anthropomorphizing AI causes people to make false predictions about what AI can do for them, what kinds of errors they make and why they make those errors. It will no doubt result in serious consequences from those unanticipated errors.
Stop anthropomorphizing LLM “thinking” and “reasoning” traces! New paper:.
0
0
7
This io / Ive / OpenAI deal is dodgy as hell. $6.5Bn for no customers and no revenue. I suppose that's the price to keep fueling the hype train these days.
0
0
0
My regular PSA for new models, this time Claude 4 Sonnet. Mistakes on simple tasks like this do not diminish on more complex tasks, they compound.
0
0
0
How this is not setting off glaring alarm bells in the industry is frankly beyond my comprehension. 48% hallucination in OpenAIs own testing! 48%!.
An AI leaderboard suggests the newest reasoning models used in chatbots are producing less accurate results because of higher hallucination rates. Experts say the problem is bigger than that
0
0
0
I can confidently say I called this one right nearly 2 years ago in a public presentation.
fastcompany.com
Three years into the boom, it looks like AI is reshaping existing jobs more than creating new ones.
0
0
0
This is one of the most disturbing things I've read in a very long time.
Zuckerberg explaining how Meta is creating personalized AI friends to supplement your real ones: “The average American has 3 friends, but has demand for 15.”
0
0
0
RT @RyanHoliday: The secret to success in almost all fields is large, uninterrupted blocks of focused time.
0
586
0
I actually hope this is true and it becomes a $4 trillion warning signal to the world about the limits of "AI".
Did the White House use ChatGPT Vibe Code a retarded formula for our tariffs and then were too dumb to cover it up?. Yes. Here's the wildest trade story you missed today 🧵👇.
0
0
0