Tag: LLM
-
Diminishing returns of LLMs doesn’t stop progress
Recent discussions about LLM diminishing returns suggest OpenAI’s next frontier model may not be significantly smarter than GPT-4. However, this plateau in intelligence doesn’t diminish the technology’s potential, as improvements can focus on making models cheaper, faster, smaller, and better at specific tasks rather than increasing raw intelligence.
-
People tend to overestimate their content value in LLM training
Mark Zuckerberg’s recent comments on AI training data sparked some debate about content value and compensation. While it’s true that individual contributions may have minimal impact on model training outcomes, the collective value of the content we create raises questions about societal compensation. However, I think that getting free access to powerful AI models serves…
-
My thoughts on the different generative AI tools I’m using
TL;DR Here are is my ranked list of suggestions, based on my own experiences and use-cases: Over the past year or so, I’ve been experimenting with a few different language models and image generators. Over time, I narrowed in on Claude, and wrote about my preference for using it over other options. A couple of…
-
Physio is a conversational AI agent to help with physical rehabilitation
Almeida, R., Sousa, H., Cunha, L. F., Guimarães, N., Campos, R., & Jorge, A. (2024). Physio: An LLM-Based Physiotherapy Advisor (arXiv:2401.01825). arXiv. http://arxiv.org/abs/2401.01825 Abstract “The capabilities of the most recent language models have increased the interest in integrating them into real-world applications. However, the fact that these models generate plausible, yet incorrect text poses a…
-
OpenChat is an open-source language model
in AIOpenChat is an open-source large language model being developed by a student team at Tsinghua University in China. In Dec 2023, OpenChat was GPT-3.5 equivalent, outperforming ChatGPT (Mar 2023) in some benchmarks i.e. the free version of ChatGPT, not ChatGPT Plus which is running GPT-4. Despite our simple approach, we are committed to developing a…
-
Follow-up prompts to generative AI responses
If your initial prompt isn’t specific enough, generative AI / Large Language Models can sometimes make assumptions about your level of understanding of the problem. In these cases, it can sometimes feel like you’ve hit a barrier because the response to your original question or instruction may not be helpful. In these cases you need…
-
AI and expertise in professional education – ADAPT conference
At the 2023 ADAPT physiotherapy conference, I discussed the impact of generative AI on physiotherapy education, emphasising how it challenges the traditional role of universities in professional training by offering accessible and personalised expertise. The presentation highlighted AI’s potential to democratise learning, enhance project-based education, and create affordable learning alternatives outside formal degree programs. I…
-
What would Claude do?
Here’s another example of a different way to think about language models: ask it for concrete advice when you’re faced with a difficult decision. In particular, the kinds of decisions that need to integrate lots of context, many variables, with high levels of uncertainty, and high-stakes outcomes. The kinds of decisions that, even with the…
-
An AI-supported health professions tutor (and a brief comparison with Claude)
https://hpe-bot.com/ Update: For more information about the additional functionality of this bot, see this LinkedIn post by Lambert Schuwirth. The interface is a bit basic, and I haven’t played around with it much, but it seems to do a decent job with the kinds of questions I asked. Also, the premium account seems to offer…
-
Claude, help me to write
Yesterday I published a post describing my concerns with how universities are responding to the new paradigm of expertise-on-demand that’s facilitated by generative AI. At the end of that post I noted that I wrote it collaboratively with Claude, and this post describes what that process (kind-of) looked like. I also want to be clear…
-
When an unstoppable force meets an immovable object
TL;DR (generated by Claude, lightly edited by me). The rise of abundant expertise in the form of generative AI questions the university monopoly on expertise provision and validation. Leadership in the creative deployment of AI for learning, teaching, and assessment will require a change in mindset and a shift towards a new paradigm, which universities…
-
Using LLMs to raise our expectations
Instead of spending so much of our time thinking about whether students 1) should use LLMs, 2) how much they use it, and 3) what they use it for, let’s just say that they should use it, a lot (IMO, the more, the better), and look more closely at what they use it for, and…
-
I tell my students to use language models to replace me
in AII know there’s a lot of concern about the potential issue of using LLMs like ChatGPT and Claude to write essays. And it’s true that there’s a risk they can be used for this. But I worry that we’re placing too much emphasis on a risk that I think largely misses the point, which is…
-
Claude, help me choose a career
Most people on the planet will need to make a series of decisions about how they are going to earn a living. And most people lack the networks and access to expertise to help inform those decisions in the best way. You might end up in a career because of social pressure, lack of opportunity,…
-
Claude, help me draft the outline of my academic paper
My last 2 posts have dealt with 1) the use of Claude to complete a peer review, and 2) how journals could include this process in their workflow. It follows that authors should be using LLMs as well. There are the obvious use cases; rephrasing passages, summarising, expanding, correcting, and so on. However, I think…
-
Journals should be experimenting with LLMs in their editorial workflow
I recently wrote a post about using Claude to peer review an academic paper, and the decent job it did. Based on that experience, I started thinking about the probable impact on journal editorial workflows, a significant part of which is the peer review process. If I was still on an editorial board of a…
-
Claude, you are an expert peer-reviewer…
I recently completed a peer review for an academic journal, and as I was submitting it I wondered how Claude would perform if given the task. Since the article was anonymised, I didn’t think there’d be any problems uploading the PDF and asking Claude to review it. And, I had already submitted my review so…
-
User guide for Claude AI
This user guide for Claude is an excellent resource, not only for understanding how you can use Claude more effectively, but for understanding language models in general. There’s an introduction, sections on prompt design and useful hacks to improve Claude’s responses, and an overview of the use cases you might consider for Claude. It also…
-
Article: CORE-GPT: Combining Open Access research and large language models for credible, trustworthy question answering
Pride, D., Cancellieri, M., & Knoth, P. (2023). CORE-GPT: Combining Open Access research and large language models for credible, trustworthy question answering (arXiv:2307.04683). arXiv. http://arxiv.org/abs/2307.04683 In this paper, we present CORE-GPT, a novel question answering platform that combines GPT-based language models and more than 32 million full-text open access scientific articles from CORE. We first…