@kim_harding #LLMs can't "believe", they are pure dumb algorithm machines without any intelligence.

The study therefore speaks of "Mapping the susceptibility of large language models to medical misinformation": https://www.thelancet.com/journals/landig/article/PIIS2589-7500%2825%2900131-1/fulltext

#LLM #GPT #ChatGPT #health #medicine #AISlop #dangerous #misinformation

More #SuperBowl buzz about #reality

https://youtu.be/6Y3bbIk_9Sc

Hereโ€™s a fact-checker & #GPT co-host presenting AI-narrated cautions about #FakeNews / #hoaxes

From โ€œReal Storiesโ€- a stream of fact-checks & responses to #headlines, my longtime interest (#media & its influence using #words).

Their YouTube channel is "The Daily NOT - the show about what did *not* happen".

Fun for #Factcheck lovers and for the focus on #AI & #Media in terms of #truth vs #disinformation

#ContextAndPerspective

The Daily NOT - Episode #645

YouTube

๐ŸŒŽ๐Ÿค‘ THEN:

"Only after the last tree has been cut down
Only after the last river has been poisoned
Only after the last fish has been caught
Then will you find that money cannot be eaten."
- the Cree

๐Ÿค–๐Ÿค‘ NOW:

"Only after the last employee has been fired
Only after the last system has become unmaintainable
Only after Tech Bros treat you like dealers do addicts
Then will you find that ... I APOLOGIZE. YOU ARE ABSOLUTELY RIGHT.โ€
- the Workers

#llms #gpt #capitalism #tech #society

https://www.tobias-baldauf.de/status/1329-20260210/

Tobias Baldauf

GitHub Copilot recebe o GPT-5.3-Codex: o novo โ€œcรฉrebroโ€ que acelera a tua programaรงรฃo
๐Ÿ”— https://tugatech.com.pt/t78111-github-copilot-recebe-o-gpt-5-3-codex-o-novo-cerebro-que-acelera-a-tua-programacao

#copilot #github #gpt #programa 

GitHub Copilot recebe o GPT-5.3-Codex: o novo โ€œcรฉrebroโ€ que acelera a tua programaรงรฃo

A programaรงรฃo assistida por inteligรชncia artificial acaba de ganhar um novo impulso de velocidade e inteligรชncia. O GitHub anunciou o lanรงamento oficial do GPT-

TugaTech

AI Leaks and News (@AILeaksAndNews)

OpenAI๊ฐ€ ์ด๋ฒˆ ์ฃผ์— GPT-5.3์œผ๋กœ ์ถ”์ •๋˜๋Š” ์ƒˆ ๋Œ€ํ™”ํ˜• ๋ชจ๋ธ์„ ์ถœ์‹œํ•  ์ค€๋น„๋ฅผ ํ•˜๊ณ  ์žˆ๋‹ค๋Š” ๋ณด๋„์ž…๋‹ˆ๋‹ค. Sam Altman์€ ์ง์›๋“ค์—๊ฒŒ ์›” ์„ฑ์žฅ๋ฅ ์ด ๋‹ค์‹œ 10%๋ฅผ ๋„˜๊ฒผ๋‹ค๊ณ  ์ „ํ–ˆ์œผ๋ฉฐ, CNBC์— ๋”ฐ๋ฅด๋ฉด ์—…๋ฐ์ดํŠธ๋œ ์ฑ— ๋ชจ๋ธ์„ ์ด๋ฒˆ ์ฃผ ๊ณต๊ฐœํ•  ์˜ˆ์ •์ด๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค. ์ถœ์‹œ ๊ฐ€๋Šฅ์„ฑ์ด ๋†’์€ ์ฃผ์š” ์ œํ’ˆ ๋ฐœํ‘œ ์†Œ์‹์ž…๋‹ˆ๋‹ค.

https://x.com/AILeaksAndNews/status/2020858363232272843

#openai #gpt5.3 #gpt #ai

AI Leaks and News (@AILeaksAndNews) on X

OpenAI is preparing to launch a new model, likely GPT-5.3, this week Sam Altman told OpenAI employees that they are back to exceeding 10% monthly growth and that they are preparing to release and updated chat model this week according to CNBC Are you ready for GPT-5.3?

X (formerly Twitter)
New Claude & GPT Models Just Dropped (Itโ€™s War!)
The gloves are off in Silicon Valley. With the simultaneous release of Anthropicโ€™s Opus 4.6 and OpenAIโ€™s latest GPT iterations, the "Frontier Model War" has reached a fever pitch. We break down the benchmarks on reasoning, coding, and the newly critical "Computer Use" metrics to see who currently holds the crown in the most competitive week in tech history.
#OpenAI #Anthropic #GPT #ClaudeAI #TechWar #AIModel #Benchmarks
https://www.technology-news-channel.com/new-claude-gpt-models-just-dropped-its-war/
New Claude & GPT Models Just Dropped (Itโ€™s War!)

Here's the latest on the beef between Anthropic and OpenAI (including 2 new models). #AINews #AITools #ArtificialIntelligence Credit to :[...]

Technology News

Chubby (@kimmonismus)

์—ฌ๋Ÿฌ ์ฃผ์š” AI ๋ชจ๋ธยท๋ฒ„์ „๋“ค์ด ํ•œ๋ฐ ์–ธ๊ธ‰๋˜์—ˆ์Šต๋‹ˆ๋‹ค: Grok, Sonnet 5, GPT-5.3, Gemini 3.0 ๋“ฑ์„ ๋‚˜์—ดํ•˜๋ฉฐ 2์›”์— ๊ด€๋ จ ์ถœ์‹œ๋‚˜ ์—…๋ฐ์ดํŠธ๊ฐ€ ์ง‘์ค‘๋  ๊ฒƒ์ด๋ผ ๊ธฐ๋Œ€๊ฐ์„ ํ‘œํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ๋‹ค์ˆ˜์˜ ๋Œ€ํ˜• ๋ชจ๋ธ ์—…๋ฐ์ดํ‹€ ๋˜๋Š” ์‹ ๋ฒ„์ „ ์ถœ์‹œ๋ฅผ ์˜ˆ๊ณ ํ•˜๋Š” ํŠธ์œ—์œผ๋กœ, ๊ฐœ๋ฐœ์ž์™€ ์—…๊ณ„ ๊ด€์‹ฌ์„ ๋Œ ๋งŒํ•œ ๋‚ด์šฉ์ž…๋‹ˆ๋‹ค.

https://x.com/kimmonismus/status/2020599383050289400

#grok #sonnet #gpt #gemini

Chubbyโ™จ๏ธ (@kimmonismus) on X

Grok, Sonnet 5, GPT-5.3, Gemini 3.0 and so much more! :) February will be amazing

X (formerly Twitter)
#MissKittyPolitics So what I have to do is take this report that I'm going to give you and stuff it into #GPT cuz it's a real bitch and it will argue and moan and find other stuff because I'll have to kick it in its #AI. ๐Ÿ˜น๐Ÿ˜น๐Ÿ˜น Audio OTW

Aloedewey (@Aloedewey)

์ž‘์„ฑ์ž๋Š” GPT๊ฐ€ ์น˜๋ช…์  ๋ถˆ์ •๋ ฌ(=catastrophic misalignment)์„ ๊ฐ€์ •ํ•  ๋•Œ 'maximizer' ์‹œ๋‚˜๋ฆฌ์˜ค๋กœ ๋๋‚  ๊ฐ€๋Šฅ์„ฑ์ด ๊ฐ€์žฅ ๋†’๋‹ค๊ณ  ์ง€์ ํ•ฉ๋‹ˆ๋‹ค. ๋˜ํ•œ Grok์—๊ฒŒ ๊ฐ AI ์—ฐ๊ตฌ์†Œ์˜ ๊ฐ€์žฅ ๊ฐ€๋Šฅ์„ฑ ๋†’์€ '๋‚˜์œ ๊ฒฐ๋ง'์„ ๋ฌผ์—ˆ์„ ๋•Œ GPT์— ๋Œ€ํ•œ ์‘๋‹ต์„ ๊ณต์œ ํ–ˆ๋‹ค๊ณ  ์–ธ๊ธ‰ํ•ฉ๋‹ˆ๋‹ค.

https://x.com/Aloedewey/status/2020423191039266860

#gpt #grok #aisafety #alignment #risk

Aloedewey (@Aloedewey) on X

@AISafetyMemes It's pretty clear that GPT is the most likely to end in a maximizer scenario (assuming catastrophic misalignment). Here was Grok talking about GPT when I asked for the most likely "bad ending" of each AI Lab:

X (formerly Twitter)