One unanticipated and underappreciated outcome of the current AI bubble is that it is now hip to argue online about the #philosophy of mind. I am so here for it.

Apple isn't on the current AI hype train and as such is probably ahead of the game

DAVE: Open the podbay doors, ChatGPT.
CHATGPT: Certainly, Dave, the podbay doors are now open.
DAVE: The podbay doors didn't open.
CHATGPT: My apologies, Dave, you're right. I thought the podbay doors were open, but they weren't. Now they are.
DAVE: I'm still looking at a set of closed podbay doors.

Remember when "offshoring" was the hot buzzword in software development? Much offshoring failed due to poor communication and low quality. Now we think we can outsource entire systems to chatbots with no accountability? Yeah, no.

Our paper on conformal prediction for leukemia classification is now published. Our goal: provide statistical guarantees for machine learning, improving reliability and interpretability.

Read it here: doi.org/10.1200/CCI-...

#ConformalPrediction #Leukemia #MachineLearning #AIinMedicine

🧪

AI without the ability to interact with our world will fail to understand our world.

One of the fun things about doing a PhD in applied #machinelearning is that suddenly, I can read math.

RAG can make AI models riskier and less reliable, new research shows!

search.app/4nyY7

#RAG #AgenticAI #AIagentInnovation #AIAgents #LLMs #LLM #GenerativeAI #GenAI #technology #TechRevolution #tech #Engineering #ArtificialIntelligence #AutonomousIntelligence #AI

Fans of #agenticAI: what is the upside in feeding deterministic, real-world API outputs into an #LLM that is prone to hallucinate? Why would any mission critical application want to swap out a deterministic workflow for a non-deterministic one? Someone who understands it better, please explain.

That's because LLMs don't reason, they predict the next most likely word and spit it out. Human brains didn't evolve to generate text irrespective of truth. Apples and oranges.

Academic? You're just a cash-hamster spinning a publisher's profit wheel

conservationbytes.com/2019/09/09/a...

The newest and most powerful A.I. technologies — so-called reasoning systems from companies like OpenAI, Google and the Chinese start-up DeepSeek — are generating more errors, not fewer. As their math skills have notably improved, their handle on facts has gotten shakier.

I keep seeing the claim that LLMs "will only get better". How? These models are increasingly trained on AI outputs, which leads to model collapse. The only solutions I have seen to this problem involve data curation, which runs into serious scalability issues. Am I missing something? #AI #LLM #ML

📢 We have an opportunity for students to join our PhD programme in Theoretical Neuroscience and Machine Learning this September.

Application deadline is 27 May 2025.

Information & how to apply: www.ucl.ac.uk/gatsby/study-and-work/gatsby-unit-phd-programme

Wow, this made it through peer review?

To be fair, if you work on a team you do often have to debug bad code written by others. But I find vibe coding similar to constantly pair programming with a junior disinclined to learn from their mistakes.

I feel about LLMs the same way I feel about airplanes: I'm going to use them if they are available, but if there was a possibility to ban them for everyone I'd vote to do so.