Comments for Tobias J. Osborne's research notes https://tjoresearchnotes.wordpress.com An open science weblog focussed on quantum information theory, condensed matter physics, and mathematical physics Fri, 13 Mar 2026 16:33:49 +0000 hourly 1 http://wordpress.com/ Comment on The Real AI Problem Isn’t the One You’re Being Sold by tobiasosborne https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21613 Fri, 13 Mar 2026 16:33:49 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21613 In reply to Vinn.

Many thanks for your comment! I am working on a series of seminars on this topic, maybe they are helpful. Take a look at https://github.com/tobiasosborne/ai-agents-seminar.git the pdf slides are in the slides subfolder. I also recommend the Vibe Coding book by Kim and Yegge

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by Vinn https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21612 Thu, 05 Mar 2026 14:23:00 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21612 Hi Tobias!

Honestly, the more I have been hearing about the current state of AI technology, the more it has had me scared for the possible kind of future it may bring. I am also a firm believer that doing your best to understand and learn about the issue is the first step for being able to help solve it, but I have been at a loss for materials on where to start to get a general image of how LLMs work. Also, I fear I’ll only be eligible to discuss the matter when I am more educated on it, so I was wondering if there are some accessible, easy-to-start-from materials which you would recommend about LLMs and/or AI as a whole?

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by aram https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21611 Wed, 25 Feb 2026 09:46:56 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21611 Hi Tobias,

I saw the announcement for your UCL talk and was sorry to be missing it! I also am not too worried about the singularity. But what makes you think these models couldn’t be compressed significantly? There is a long history of algorithm improvements where we get the same answers with less work. I would expect some of that to apply to LLMs. And I have no confidence about the limits. They could stop improving tomorrow or they could have a long way to go. What makes you think that we shouldn’t expect dramatic efficiency improvments in LLMs?

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by nekomusume.daisuki https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21610 Sun, 18 Jan 2026 20:39:08 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21610 In reply to tobiasosborne.

I hadn’t seen *that* video before, but I *am* on the Spectrum as well. It also often goes hand-in-hand with Pragmatic Inference Disorder…
People who have aphantasia rely on language to be accurate, because they think in words, not images. So inferring meaning from seemingly separate remarks is a challenge.
(I *kinda* learned myself to recognise these moments and deduce the inferred meaning, my son, also ASD, has problems with PID still!)

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by tobiasosborne https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21609 Sun, 18 Jan 2026 08:55:41 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21609 In reply to festivalsilly88f92b8973.

Sam has a video about it: https://youtu.be/5KebPocVj0Q?si=gzAQYdM8PCaFQcRs

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by tobiasosborne https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21608 Sun, 18 Jan 2026 08:54:54 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21608 In reply to Nekomusume Daiskuki.

Many thanks for your comment, I appreciate it! This is a great idea: I have several acquaintances who are aphantasic; I will be certain to let them know. What a wonderful idea.

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by festivalsilly88f92b8973 https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21607 Sun, 18 Jan 2026 08:48:51 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21607 In reply to Nekomusume Daiskuki.

And for anyone that reads this and goes “wait, ‘picture this’ is a figure of speech, right?”, no for most people it is literal… Just look up ‘aphantasia’, or ask your favourite LLM to tell you about it…

]]>
Comment on The Real AI Problem Isn’t the One You’re Being Sold by Nekomusume Daiskuki https://tjoresearchnotes.wordpress.com/2026/01/12/the-real-ai-problem-isnt-the-one-youre-being-sold/#comment-21606 Sun, 18 Jan 2026 08:44:17 +0000 http://tjoresearchnotes.wordpress.com/?p=872#comment-21606 I am aphantasic. AI helps me visualise what I am writing. I can describe what is happening, the environment, but I cannot see it… (Think of it like a movie where the screen is black, the speakers off, and only the subtitles are runnning)
AI images give me an approximation of what readers would see when they read it…

]]>
Comment on The Adversarial Prover: A Skeptic’s Approach to LLM-Assisted Mathematics by 仅提示的对抗式LLM代理定理证明 - 偏执的码农 https://tjoresearchnotes.wordpress.com/2026/01/02/the-adversarial-prover-a-skeptics-approach-to-llm-assisted-mathematics/#comment-21605 Fri, 02 Jan 2026 20:22:58 +0000 http://tjoresearchnotes.wordpress.com/?p=850#comment-21605 […] 详情参考 […]

]]>
Comment on The Adversarial Prover: A Skeptic’s Approach to LLM-Assisted Mathematics by AI Agent 每日资讯 – 2026年01月02日 – User's blog https://tjoresearchnotes.wordpress.com/2026/01/02/the-adversarial-prover-a-skeptics-approach-to-llm-assisted-mathematics/#comment-21604 Fri, 02 Jan 2026 16:25:12 +0000 http://tjoresearchnotes.wordpress.com/?p=850#comment-21604 […] Prompt-only theorem proving with adversarial LLM agents16:08 […]

]]>