Prof Gina Neff

Minderoo Centre for Technology and Democracy @MCTDCambridge · Jun 14, 2023

We are delighted to have joined @responsibleaiuk.

Announced at #LTW23 by @NorwichChloe @SciTechgovuk, this £31M consortium will create a UK + international research + innovation ecosystem for responsible + trustworthy AI.

@Cambridge_Uni @ginasue

mctd.ac.uk/cambridge-acad…

🧵1/6

View quoted post on X

Martha Dacombe @martha_dacombe · Apr 28

3/HMG knows Britain can't build the whole stack. But it can be indispensable at critical points — frontier research, chip design, safety, compute. Sovereignty is trying to find critical points to leverage.

View quoted post on X

Nav Toor Verified account @heynavtoor · Mar 6

🚨BREAKING: OpenAI published a paper proving that ChatGPT will always make things up.

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

Every time ChatGPT gives you an answer, ask yourself: is this real, or is it just a confident guess?

View quoted post on X

Rapporteur @rapporteur_news · Feb 3

TOMORROW | AI & disinformation – How can Europe safeguard trust in the media?

Join to discuss how AI can strengthen the EU’s response to disinformation while safeguarding media freedom & trust.

Register: eurac.tv/9Ykd
More info: eurac.tv/9Ymb

#ai4trust

View quoted post on X

derek guy Verified account @dieworkwear · Sep 3, 2025

very few ppl are as talented and fortunate as bill cunningham. but i often think about how he was able to live an honest life doing what he loved bc he lived simply. even at the height of his career, he slept on cot. "if you don’t take money, they can’t tell you what to do."

View quoted post on X