https://stepsandleaps.wordpress.com/feed

In the previous post I talked about an impressive creative application of LLMs to a complicated problem in high-energy physics. One interesting concept that came out of it was that a human expert still had to decide which research directions were likely to be the most profitable and this was called ‘taste’. Perhaps it could also be called an aspect of ‘wisdom’.

Independently I’ve been curious what the leaders in AI have been thinking on how the area would progress and how it related to wisdom as well as knowledge. Each company will obviously have expert teams working on the funding/finance issues and also on the large-scale technology (data centres/software) developments. However the leadership, apart from these key business issues, must also be giving time to bigger picture societal thinking – where is this all going? I was curious who helps them with this and what their conversations were like.

So I was fascinated when I recently came across a (38 min) discussion between Joe Hudson and Brett Kistler, which gives a small flavour of this. It’s an indirect advert for his coaching business but it does contain lots of interesting big picture ideas and stories.

“Joe coaches the founders of OpenAI and works with people from Anthropic and DeepMind. This (session) is an episode about what changes when intelligence gets outsourced: the decisions you make, the relationships you hold, and whether you can do the hard thing when it matters. Joe breaks down the 3 games – outer, inner, and team – that determine who thrives in this era, and they end with practical ways to tell if your AI use is making you more connected or less.”

Some key takeaways from the video were:

• The key role of human wisdom versus outsourced intelligence/knowledge.
• The ability to make (good) hard decisions. Can you and your team have hard conversations? How can you create alignment within conflict?
• There is likely to be societal rupture through AI, with a period of turmoil and dysfunction to change values.
• There is the opportunity for something horrible and for something great (fairly obvious of course and likely both will happen).
• Heightened role of excellent (small) team work, so how to build close and successful teams? How can a team rise above itself? This is a timeless question but now with more consequence.
• Feeling of a team as a family (empathy, connection, collective – being ‘human’).
• We are raising AI and AI is raising us.

A tip was to make sure you’re talking to real humans about the hard stuff and find a way to do this, it shouldn’t just be random.

If you want to just dip in, the chapters are:

0:00 Introduction
0:40 Joe coaches AI founders
2:35 Wisdom is being good at being human
5:49 The opportunity to just focus on being human
8:45 The fear of losing your job is really about worth
13:57 Three games that determine who thrives
16:00 AI as the great equalizer
23:38 We are raising AI and it’s raising us
28:23 Your attachment style shows up with AI too
33:10 Talk to a human about the hard stuff
37:25 It can get better than you think

I spent the first 10 years of my career doing academic research in theoretical physics (I then left for hi-tech). However that investment in knowledge has never left me and I try to keep up, at least in general terms.

One of the main books in the area I worked in (that I still dip into now and again) is written by Professor Matthew Schwartz of Harvard University (see here). So I was interested to find that he had written a post on the Anthropic blog on how he creatively used AI as an assistant in his research: Vibe physics: The AI grad student.

This was especially instructive as I have written earlier (here) on another attempt to use AI in theoretical physics that didn’t go so well and prompted a lively discussion.

Schwartz’s final research paper is obviously extremely technical (preprint here) and the details will only appeal to specialists but the blog post is very useful as he summarises general lessons that have a wider context.

Here are some extracts:

“What about theoretical physics? End-to-end AI scientists have found their footing in data-rich domains, but theoretical physics is not one of them. Unlike mathematics, theoretical physics problems can be more nebulous—less about formal proof search and more about physical intuition, choosing the right approximations, and navigating a landscape of subtleties that often trip up even experienced researchers. Even so, there are problems in physics where AI might be better suited. Not yet the paradigm-shifting questions at the frontier, but those where the conceptual framework is established and the goal well-defined…

I picked this problem because it connects directly to the foundations of our understanding of quantum theory. But more importantly, it’s a highly technical calculation that I was confident I could do myself. The physics is understood in principle; what’s missing is a careful, complete treatment…

Yet, if this had been the first project I did with a graduate student, I would also have had to check everything, so maybe this is not so surprising. But a graduate student would never have handed me a complete draft after three days and told me it was perfect….

It took about a week to get the results right. I had Claude write out all the details of every calculation—in much more detail than had been included in the paper—and had GPT and Gemini check those calculations first. If all three agreed, it was a good indication it was correct. Even so, I went through and discovered a few examples where all three missed some terms…”

These were calculational and presentation issues, that could be checked by an expert(s) but a wider point I found the most fascinating:

“I am more confident that the bottleneck is not creativity. LLMs are profoundly creative. They simply lack a sense of which paths might be fruitful before walking them. I think we can distill what is missing in current LLMs to a single word: Taste.

In physics, taste is the intangible sense about which research directions might lead somewhere. I’ve been doing research in theoretical physics for a long time and have learned to tell pretty quickly whether an idea is promising or not. I suspect anyone who has honed a craft for a long time—whether in science, carpentry, or design—would recognize this: experience produces a kind of judgment that AI has not yet mastered. We do not give enough credit to taste. When solving problems is hard, the solution gets the glory, but when knowledge and technical strength are ubiquitous, it’s the taste to come up with good ideas that distinguishes great work.”

This last point on Taste leads on nicely to the following post on Being Human, Wisdom and AI.

Aside: If anyone is interested in this area and would like a book to explain the basics descriptively then Quantum Field Theory, As Simply As Possible by Anthony Zee is a good start.

We are often quiet and private about our life struggles (whatever they may be) but delighted to talk endlessly of our achievements and interests leading to a very imbalanced view of everyone’s life.

So it’s instructive and revealing when someone who you might think ‘has it all’ is honest and describes how deep negative moods can develop but how these can also spontaneously disappear, often through simple triggers.

As an example, Brian Eno (musician) writes (my emphasis in bold):

“Experience has shown me that, when I reach this point, all the distractions I can muster are only postponements. It’s time to face up to total, unmitigated despair.

I sometimes do this by going alone on a “holiday” — though that word scarcely conveys the crashing tedium involved, for I usually choose somewhere uneventful, take nothing with me, and then rely on the horror of my own company to drive me rapidly to the edge of the abyss….

It goes like this: me thinking, “What’s it all for?/ What’s the bloody point?/ I haven’t done anything I like and I don’t have a clue what to do next/ I’m a completely empty shell.” This lasts two days or so… Then I suddenly notice — apropos of something very minor, like the way a plane crosses the sky, or the smell of trees, or the light in the early evening, or remembering one of my brother’s jokes — that I am thoroughly enjoying myself and completely, utterly glad to be alive. Not one of the questions I asked myself has been answered. Instead, like all good philosophical questions, they’ve just ceased to matter.”

He then went on to a very productive and creative period.

Sometimes you have to get away from everything to get your life in perspective. Thinking about things alone or talking with friends may simply not be enough.

For less extreme situations, there’s a recent article by Anne-Laure Le Cunff on why rest alone doesn’t always lead to a feeling of restoration and energy. She is both a research scientist and an entrepreneur. Based on her neuroscience expertise, she’s a strong advocate of an ‘experimental’ mindset and her Ness Labs site has a large and free knowledge base of articles on this topic. I like that she categorises these under both Best and Most Popular!

A few months ago I picked up a book on the Camino de Santiago from the local library, not because I’m religious but because I liked the graphics, photos and presentation. In fact I later tracked down the illustrator, Claire Harrup, and bought one of her prints from her website (see above).

Because of this and even though I have banned myself from buying more books I weakened and decided to buy Walking the Camino de Santiago as a fond memory.

I was toying with buying it online but then decided to support the bookstore Waterstones instead, mainly because their staff are always friendly, chatty and helpful. The difference in price is easily compensated for by the delightful service.

I was browsing the book, prior to buying it, as they had it in store. One of the assistants was passing by and asked whether I was planning on travelling the Camino de Santiago. I said no it was mainly for the graphics but through it I had discovered that the north of Spain looked really interesting and not touristy. Rather amazingly he said that he used to live there, in the Leon Province. He then gave a fascinating summary of the history of the Camino from pre-Christian times. Then some history of why that strip of Spain was so important, the role of the Moors, the Templars and consequently the numerous large forts and churches. Separately (and just as interestingly) he also gave a background on why he lived there for 10 years (I did ask). A fascinating, concise and big picture summary. As a consequence of this I’m going to investigate this topic further as my curiosity was piqued (plus probably buy a few more books).

All this from a simple question. I was thinking of the contrast to an online transaction where price is the only benefit.

The conversation made my day, I was totally fascinated. It also reminded me that in these increasingly digital days how much we humans have to offer. It was a good example of serendipity and someone taking the initiative by asking a simple question. Something we should all do more of, both at work and with friends and also strangers. You may be surprised and delighted with the results.

This topic reminded me of a previous post of mine on how friends and parents often have little clear idea of what those close to them actually do, apart from bland generalities.

Finally I wish to emphasise that I am in no way related to or involved with Waterstones. They’re just a great local bookstore.

Image: The is the best graphic I’ve seen (in French but easy to guess meanings) as it shows a few more steps. It’s still a simplification of course.

I started my career as a research academic (theoretical physics) and worked in this area for over 10 years. After a career jump, I held positions at an IBM research centre, a mathematical modelling group in a government agency and finally leading consulting in innovation for a large commercial company. This corresponds to going in stages from idea generation and development in small teams, to scaling things up (often in national collaborations) to finally implementing initiatives at an international level. It also spans a wide range of underlying cultures.

As organisations became more complicated and there are more options, you realise there are many ways of progressing innovation, most of which are far from perfect. Apart from the necessity of germinal ideas in a dynamic and complex landscape, equal attention has to be paid to obtaining sustainable funding which naturally encompasses company politics and top-team support.

A diverse balance of business skills is required, just having a bright idea is certainly not enough. I always thought there existed a valuable role for a group of people with special skills that could help overcome common obstacles.

In this context I was interested to read two articles that discuss approaches to this problem that mirror my own experiences and views. The key idea is to have people that assist innovation by their depth and width (’T-Shaped People’) and/or emotional and contextual intelligence (‘Bridgers’).

From Heike Riel, IBM Fellow and Head of Science and Technology at IBM Research, in the CERN Courier (my emphasis in bold and this links to the graphic above):

“Riel champions the idea of the “T-shaped person”: someone with deep expertise in one area (the vertical stroke of the T) and broad knowledge across fields (the horizontal bar of the T). “You start by going deep – becoming the go-to person for something,” says Riel. This deep knowledge builds your credibility in your desired field: you become the expert. But after that, you need to broaden your scope and understanding.

That breadth can include moving between fields, working on interdisciplinary projects, or applying physics in new domains. “A T-shaped person brings something unique to every conversation,” adds Riel. “You’re able to connect dots that others might not even see, and that’s where a lot of innovation happens.”

Adding the bar on the T means that you can move fluidly between different fields, including through academia and industry.”

She also makes a side-point I hadn’t appreciated as had taken for granted, that through academic training you get used to and are not phased by failure but rather regard it as a natural step on the journey:

“On top of that, research in physics increases your “frustration tolerance”. Every physicist has faced failure at one point during their academic career. But their determination to persevere is what makes them resilient. Whether this is through constantly thinking on your feet, or coming up with new solutions to the same problems, this resilience is what can make a physicist’s application pierce through the others. “In physics, you face problems every day that don’t have easy answers, and you learn how to deal with that,” explains Riel.”

The key, she says, is to stay flexible, curious and grounded in your foundations. “Build your depth, then your breadth. Don’t be afraid of crossing boundaries. That’s where the most exciting work happens.”

In this same vein there’s an interesting article ‘Why Great Innovations Fail To Scale’ by Linda A. Hill, Emily Tedards and Jason Wild in the Harvard Business Review. This emphasises the need for emotional and contextual intelligence.

“Scaling innovation today demands contributions from multiple partners. Many innovations fail not because of flawed ideas but because teams and organizations struggle to collaborate across boundaries. What’s needed is a particular kind of leader: the bridger. Bridgers excel at curating the right partners, translating across their differing ways of working, and integrating their efforts to maintain momentum. Bridgers’ effectiveness stems from both emotional intelligence and contextual intelligence. They understand each stakeholder’s environment, pressures, and values and know how to adapt from one context to another.”

The article in the HBR is adapted from their wider-ranging and interesting book Genius at Scale (Harvard Business Review Press, 2026).

It’s interesting that innovation is such an open, fascinating and never-ending story with new angles and perspectives constantly coming out. The challenge is to find what works for you, be it large company or SME.

Scott Berkun has written some interesting books. He’s now writing (in public) a book on the popular topic of Rules For Life. I like his honesty:

“Socrates was ignorant of the fact that many people are happy with life without much examination. And I’ve written before about how we admire cats and dogs for their simple lives, unburdened by the need to question everything. I also believe that feeling more, and being more in touch with our emotions, might be a better path to fulfilment than relentless intellectual examination of things.

However for this project I must scrutinize, examine, provoke and overthink to produce a book worthy of a title like “Rules to Live by.” I will keep plugging away!”

Browsing in a bookshop, I was fascinated to find there was a book on the topic of philosophy and dogs. The front cover says it all

“Drawing on his life lived with dogs (two German shepherds, the amiable Hugo and his dark twin Shadow; Brenin, a wolf hybrid, and Tess his wolf dog daughter; and Nina, a German shepherd/malamute mix), on the ideas of philosophers from Socrates to Hume and Sartre, and on the cutting edge psychology of canine cognition, philosopher Mark Rowlands explores the way dogs experience the world to bring us closer to an understanding of ourselves.

While dogs feel unparalleled joy and focus in the moment, humans are burdened by the disquietude of anxiety, doubt and even anguish. Happiness for dogs can be achieved in the daily chase of a squirrel, for humans it is much more elusive. Digging deep into their morality, freedoms, consciousness, intelligence and love of life, Rowlands discovers that dogs have a unique way of existing which amounts to a different philosophical outlook altogether – if they could write such a thing – and that they may have better answers to the meaning of life than we do.”

I’ve just loaned it from the local library to skim. Dipping in and out it seems an amusing and insightful read. One aspect is that a dog’s life is full of spontaneity whilst for humans it’s often dominated by planning (weekends, events, holidays, getting together with friends etc) that often excludes spontaneity. So we can all try to be a bit more spontaneous!

Last week there was a exceptionally clear popular overview talk (50 mins + Q&A) on AI and LLMs by Professor Michael John Wooldridge of Oxford University. He was awarded the prestigious Royal Society Michael Faraday Prize for 2025 and this was the accompanying lecture:

“The Royal Society Michael Faraday Prize and Lecture is awarded annually to the scientist or engineer whose expertise in communicating scientific ideas in lay terms is exemplary. The award is named after Michael Faraday FRS, the influential inventor and electrical pioneer who was prominent in the public communication of science and founded the Christmas lectures at the Royal Institution”

“In his talk, Professor Michael John Wooldridge will look at how the new AI works and why, as a consequence, it exhibits these weird, frustrating, fascinating behaviours. He will show just how far the new AI is from classical expectations and talk about the next frontiers for AI – and how far we are from the dream.”

There are also some interesting insights from him in a recent article in the Guardian including:

“The risk is that people start treating AIs as if they were human. In a 2025 survey by the Center for Democracy and Technology, nearly a third of students reported that they or a friend had had a romantic relationship with an AI.

“Companies want to present AIs in a very human-like way, but I think that is a very dangerous path to take,” Wooldridge said. “We need to understand that these are just glorified spreadsheets, they are tools and nothing more than that.”

Wooldridge sees positives in the kind of AI depicted in the early years of Star Trek. In one 1968 episode, The Day of the Dove, Mr Spock quizzes the Enterprise’s computer only to be told in a distinctly non-human voice that it has insufficient data to answer. “That’s not what we get. We get an overconfident AI that says: yes, here’s the answer,” he said. “Maybe we need AIs to talk to us in the voice of the Star Trek computer. You would never believe it was a human being.””

I thought the last sentence instructive as it emphasises the difference between an impressive tool and a real-world knowledgeable friend (who could candidly admit to being wrong or misleading).

Linked to this general topic, a year ago I came across a very nice explanation of the fundamentals of AI and LLMs by Bertrand Serlet (who was Senior Vice President of Software Engineering at Apple). It goes into a bit more detail eg describes the gradient descent method for optimisation. Some basic maths is involved but not much and in my opinion the general ideas come over very clearly.

This post was triggered by the thought that I once did research based on Hopfield’s groundbreaking work on AI. This research took place at the novel multi-disciplinary IBM Scientific Centre in Winchester (now closed down). It’s amazing to see how AI has progressed since then after a number of important progress ups and downs. For his fundamental contributions to the area, John Hopfield was co-awarded the Nobel Prize in Physics for 2024.

“When we talk about artificial intelligence, we often mean machine learning using artificial neural networks. This technology was originally inspired by the structure of the brain. In an artificial neural network, the brain’s neurons are represented by nodes that have different values. In 1982, John Hopfield invented a network that uses a method for saving and recreating patterns. He found inspiration in physics’ models of how many small parts in a system affect the system as a whole. The invention became important in, for example, image analysis.”

There’s a popular exposition of the foundational AI work of John Hopfield and Geoffrey Hinton here.

Rather randomly, browsing the ever-interesting Kottke site, I came across a fun and captivating student project on constructing an autonomous flying umbrella. This might sound a bit niche and odd but actually illustrates in an entertaining way many of the key features of modern innovative research projects. The video has been cleverly edited to capture the energy of the project and its illuminating ups and downs.

I’ve written before on the challenges of creative projects (images from here):

For me the key general points from the video were:

• Research project – so exploring the ‘unknown’ within the constraints of time, interest and money
• Inspirational, very honest with the ups-and-downs
• Illustrates the strong element of creativity and problem solving

On a more detailed level it illustrates:

• Importance of pulling in extra resources and asking for help (which many are loathe to do)
• Procrastination, maybe also an element of subconscious reflection
• Making use of smart tools (3D printer, Raspberry Pi etc)
• Not Giving Up, an idea gets under your skin!
• Sometimes you have to ignore doubts and just carry on
• Letting go of perfectionism (adopt a ‘good enough’ philosophy)
• Communicating clearly on technical matters is an essential and worthwhile skill, not an optional add-on
• Commercial awareness, making deals with equipment providers

It’s an excellent educational video as it gives an exciting flavour of the issues you encounter doing technical research in the real world.

I was thinking the other day why is it that some politician’s arguments and messages are so superficially appealing. You know they’re wrong as ridiculously simplistic or even sometimes completely baffling. As well as the content, the argument is given at a speed that’s hard to cope with (perhaps for them too).

Real problems are messy, complicated and legacy-driven (which I appreciate no one wants to hear about). It takes time.

All this was neatly crystallised in a recent post by Seth Godin on fake news and trust:

“The irony is that it’s easier to trust fake news. It’s consistent, simplified, coherent and predictable, all the things that humans look for when we’re seeking solace.”

On this same theme, from Om Malik:

“That’s why we get all our information as memes. The meme has become the metastory, the layer where meaning is carried. You don’t need to read the thing; you just need the gist, compressed and passed along in a sentence, an image, or a joke. It has taken the role of the headline. The machine accelerates this dynamic. It demands constant material; stop feeding it and the whole structure shakes. The point of the internet now is mostly to hook attention and push it toward commerce, to keep the engine running. Anyone can get their cut…

The cost of all this isn’t abstract. It’s the review that took three months but no one will read. It’s the investigation that required patience. It’s the work of understanding something before declaring judgment. All of it still exists, still gets made. It just doesn’t travel. And in a system where only what travels matters, we’ve made expertise indistinguishable from noise.”

I read an interesting article on the BBC site on the important issue of widespread and rising obesity levels in the UK. Here’s an extract:

“Part of this is down to the sheer volume – and affordability – of poor quality, high-calorie foods, and in particular ultra-processed foods. Add to that aggressive marketing and advertising of fast food and sugary drinks, growing portion sizes, and limited opportunities for physical activity (often due to urban design or time pressures), and it makes for a perfect storm.”

It’s funny but I thought you could easily replace ‘food’ by ‘information/social media’, leading to ‘knowledge obesity’. All of us gorging on trivia or wild exaggeration (at least some of the time).