About four years ago I started paying attention to MDPI on the back of starting to look at Walt Crawfords amazing annual Open Access reports. (https://waltcrawford.name/goaj.html). The rate of revenue growth was astonishing, and over the next few years it continues. I truly thought at the time that maybe the existing journal model was ...

Ethan Mollick remains a key writer, giving perspective on what to expect in 2024: https://www.oneusefulthing.org/p/signs-and-portents • GenAI is already positively impacting work, with significant productivity gains (improvements of tens of percentage points). • Interestingly, many companies are either ignoring it or trying to use it a...

Here are some things that I read over the last few weeks that I found interesting, or that I learned something from. Benedict Evan’s latest newsletter: https://newsletters.feedbinusercontent.com/de8/de8e11d8b69e568a52963c89d86e8b3d3ed837c9.html](https://newsletters.feedbinusercontent.com/de8/de8e11d8b69e568a52963c89d86e8b3d3ed837c9.htm...

From Simon Willison https://simonwillison.net/2023/Dec/19/facebook-is-being-overrun/ I picked up the link to this article: https://www.404media.co/facebook-is-being-overrun-with-stolen-ai-generated-images-that-people-think-are-real/ People are using image generation tools to farm the generation of 1000s of fake variants of mildly viral...

Cross posting this blog - https://towardsdatascience.com/classifying-source-code-using-llms-what-and-how-f04c7dbcba9b It’s chock full of great advice on using LLMs. The task they worked on was determining if some given code was malicious. Some key advice: - running an LLM is expensive, check if another method might work for you. - LLMs...

I had to wrap up my climbing year a bit early. On Thursday, just over a week ago, I injured my left ring finger’s FDP (long finger flexor tendon). It’s a moderate injury; recovery should take six to ten weeks, maybe less, but it’s serious enough for me to stop climbing ahead of our Christmas trip to go skiing. In spite of this setback ...

This is a great list of factoids. https://snippet.finance/52-snippets-from-2023/ Here are my favourites, and what I think about them. Global inequality is reducing. This goes against my intuition, which tells me I remain have strong biases about how the world works and the current state of the world 37% of the worlds population has nev...

Some interesting LLM papers This is my take on papers from Davis Blaclock (https://dblalock.substack.com). This paper uses an LLM to rank response from a fine tuning run on a small model (7B parameters) to get it to perform in a more aligned way than a large model (70B) that has been fine tuned using RLHF. This is important because LLM...

Anastasia Toynbee from the Royal Society of Chemistry gave a great presentation a few weeks ago on how RSC are using machine translation to help authors who have English as a second language. There is a huge language burden for ESL authors, and she covered some of the stark data in her talk. Remember, talent is evenly distributed in th...

This week our data and technology department had a workshop looking at how to be more decisive, led by the fantastic Claire Holmes. Going away from the meeting we have asked each team to pick two techniques to apply over the next 12 weeks, and then to report back on whether they were effective. It will be interesting to see where we la...

Episode link https://www.bbc.co.uk/programmes/m001qdx1 Episode rating B+/A- I was fairly familiar with Einstein’s life, having had a fascination with the birth of modern physics for some time, but even still I learnt a few new things about him from this episode. I’d forgotten quite how long he had worked in obscurity as a parent attorn...

Episode link https://www.bbc.co.uk/programmes/m001jst2 My recollected rating: A I really enjoyed this episode, and basically learnt all about the epic, but I have now mostly forgotten the details. I Got GPT to summarise the work again for me, and what I recall from the episode is a good analysis of the role and tribulations of Sita in ...

Episode link - https://www.bbc.co.uk/programmes/m001hx7n My rating - A- The chartists were a social movement in England that ran from around the 1830s to the late 1860s. It was a mass movement in search of political reform that would rebalance power in favour of the masses. As such it can be described as a social movement. Gatherings o...

Episode link: https://www.bbc.co.uk/programmes/m001jc68 My rating: B- I learnt hardly anything new in this episode. The book - the man who loved only numbers - https://www.amazon.co.uk/Man-Who-Loved-Only-Numbers/dp/1857028295 is a much much better guide to the man. Having read that book I was excited about this episode, and was saving ...

I asked for a summary of the following paper: https://www.embopress.org/doi/full/10.15252/embr.201949472 ( The troubles with peer review for allocating research funding ) The command I used was: llm -m gpt-4-turbo -s "write an abstract for this paper, and give me the key findings in 10 bullet points" < funding-revew.txt With this comma...

Earlier this week I gave a talk about some of my current thinking about generative AI. One of the points I made is that while these tools have tremendous power, we are still scrambling around looking for the best ways to invoke that power. This reminded me of the command line. With the command line, you need to know the right incantati...

OpenAI made a ton of announcements earlier this week. As Benedict Evans pointed out, they are driving hard for the platform play. As expected, their models are becoming faster, more capable, and cheaper, with much longer context windows to boot. I fully expect this drive to continue for quite a while, as there remains a ton of headroom...

This article in The Verge is a good overview of responses from large tech companies to the US copyright office on questions around what the future of copyright should be in terms of regulation of AIs use of copyrighted material. The responses are very predictable. Mostly they argue that there should be no limits on what these AIs shoul...

I attended FuturePub last night, actually, I also spoke at it too. I love these events, I've attended a ton of them over the years, and last night's was in association with AI fringe, so there was a nice ad-mixture of different communities, and I got to chat to some folk that I wouldn't otherwise have met. A really good event, many tha...

On Monday the 30th of October I’m taking part in a FuturePub - AI and Research event. I’m going to be leading an open discussion on the topic of implementing AI Governance. EventBrite page with some more details. I put in the proposal when I saw the event pop up for a few reasons, the most direct one being that at BMJ we have implement...

I follow a monthly roundup of Machine Learning papers by Davis Blalock. https://dblalock.github.io/about/. It's a fantastic way to keep up with where the focus of the research community is, even at a very high level. You can catch it here: https://dblalock.substack.com/ I'd say the trends I've been seeing tend to be around areas such a...

Last week I took part in an internal discussion with the CIO Carolyn Brown from the BMA where we talked all things AI, and how we think about its possible impact on the workplace. I think we had about 120 folk listen in, and we recorded the session in case anyone wants to catchup. The discussion was nicely moderated by Gordon Fletcher,...

Gary Marcus writes persuasively that markets are massively over pricing LLMs, and this could lead to some very bad decision making in the near term - https://garymarcus.substack.com/p/what-if-generative-ai-turned-out. He writes that if the US gets into an AI “war” with China things could get hairy, But what has me worried right now is ...

I felt this blog post deeply - https://lethain.com/frameworks-decision-making/, as I often find myself making that internal calculation around where I am choosing to direct my energies and how well aligned my current pointer of attention is to company needs. As a manager I have been trying harder over the last few years to listen more,...

I joined in 2007, my then colleague Gavin had returned from SXSW and was talking about this new thing. It has been enormously beneficial for my career, and I’ve learnt and laughed a lot from the interactions I had on that platform. I was fortunate to never see the dark side of the web on that platform, I have a very privileged position...

A few weeks back I sat down for about an hour to be interviewed by Mike Green, whose podcast "Understand Users" is all about UX and user research - https://understandingusers.podbean.com Mike and I go back a few years, and I've seen him flourish in that career over the last few years. It was a real pleasure sharing some thoughts and re...

I’ve been sharing some links around internally at work over the last few months, so I wanted to post them here. Clearly dominated by LLMs Douglas Hofstadter has changed his mind about LLMs, and his perspective is quite sobering. I don’t give much time to folk who are very negative about the future prospects of these tools, but he is so...

https://julian.digital/2023/07/06/multi-layered-calendars/ Is a fantastic post reflecting on the poverty of our calendar applications. The author has sone fantastic suggestions on how to enhance what a calendar can be. I think that the problem with this approach is one of customisation. The product market fit for such an imagined produ...

How likely is it to generate a copyrighted image using a Generative ai tool such a stable diffusion? This question came up in a work context last week and I've been thinking about it. I asked GPT how these tools work, and it came back with "generative image models like Stable Diffusion learn from data to generate new images by applying...

This tweet thread is worth looking at: https://twitter.com/random_walker/status/1681748271163912194 What this tells us, and what is worth paying attention to, is that when building on top of LLMs that are outside of your control, the fine-tuning, or model, can move underneath you. If you have strategies for making the LLM work for you ...