As I have noted in response to comments elsewhere on this blog, I was in Wales last week, on something that some people call a holiday, but which I prefer to think of as a change of situation. That is very largely because it would appear that wherever I am, my brain refuses to stop creating new ideas at a rate faster than I can use them here, on YouTube, or in the National.
The simple fact is that, much as I enjoy being away, and spending more time every day as a consequence doing what I normally think of as weekend activities, I rarely, if ever, take weekends off, and as a result, nor am I very good at stopping on holiday.
However, there was a very strange aspect to this holiday which made it quite different to any other I can recall since I was probably a teenager. What I realised during it was that on my return, there was not going to be an overwhelming pile of other tasks to undertake, into which I would also need to try to fit blogging, making videos, and moderating comments. Instead, it finally dawned on me that those activities are now what I do.
That said, I did, with colleagues at Copenhagen Business School, have a paper accepted by a four-star accounting journal just before I went away, and I have a tiny amount of checking on that to do, now I am back, just to keep them happy.
I also have a little work to do on the project on tax transparency that I have been working on for some time, which looks as though it might now be used as training material in multiple developing countries on at least three continents.
I've also volunteered to do a little more on the Accounting Streams project, but add all these things together, and they are, at most, minor distractions and not major tasks. I even have most of my own accounting up-to-date, whilst the pile of admin that I had to deal with, having decided to end paid employment and my grant-based research activities, appears to have largely come to an end.
The result is that I generated a whole pile of ideas whilst away, many of which have a good chance of seeing the light of day, if I want them to do so.
One of those opportunities is to write a book, and there are now multiple publishers interested in me doing so. If others want to get in touch, please feel free to do so.
Another is to develop my YouTube channel, somewhat, as I mention in today's video. There are masses of ideas there, because I am finding this whole process of social media publishing absolutely fascinating.
But that then led me in another direction, which was to realise that I really do need to properly understand AI, and how to use it. I have, as a result, been reading quite a lot about this, and appreciate that even so, I am only scratching the surface of this issue.
I also admit that I have had to think about the ethics of all this, but realised when doing so that choosing to ignore AI now would be like an accountant choosing to overlook the spreadsheet revolution that began in the early 1980s, and which transformed the way in which almost all accounting departments, and accounting and business decision-making, took place. I embraced this change from the outset and acquired my first computer capable of running Lotus 1-2-3 in 1983 and never looked back. In the end, I wrote accounting packages based on spreadsheets and sold them. Whilst I would not do the same thing now, when I did, they revolutionised the productivity of my firm and the quality of its work. AI, I now realise, has the same capacity to revolutionise how I work if I correctly understand it.
The best of many videos that I watched on this issue last week (which I cannot find again right now), featured a French women who suggested that, as was the case with spreadsheets, one percent will really take advantage of the opportunities that this technology creates and put it to really constructive use, whilst 99% will, at most, use it to slightly alter their existing workflow. This latter group might do this by, for example, using AI as a slightly more advanced search engine on the internet, or to generate highly standardised email responses, which will remove from them the requirement to engage in any thought when undertaking these processes.
The key point was the last one. You can either use AI to give up thinking to the greatest possible degree, or you can use it to liberate your thinking. That is a choice that everyone has available to them.
It is my intention to use AI to liberate my thinking. I cannot think of any good reason why I should not, even taking into consideration climate issues. I am more than willing to give up other forms of consumption to compensate for this.
This, however, requires an intelligent use of the technology. Some of what we are already doing uses AI. For example, we use AI to correct small mistakes in videos so that we do not have to re-record if they occur. A voice clone within our software can do that, and the time saving is enormous.
We also use AI to now analyse a video's performance, and we are planning to keep doing so because it appears capable of suggesting content that will appeal to audiences, and ways to write thumbnails and titles that work on YouTube, which is a decided skill. We use dedicated AI tools for this purpose, and it seems to be working.
More specifically, I have recently added some references to the bottom of some posts, suggesting similarly themed posts that might be of interest on the blog to the one it has just been posted on. Plenty of other media have been doing this for a while, and it would appear to make sense to do so here. However, as I have yet to define the prompt for this entirely appropriately, I will not regularly use it until I have. Comments on whether this is useful or not would be appreciated. The time involved will be very limited if I get the prompt right, but these things are quite complicated. In particular, this one requires an instruction to ChatGPT to write the necessary listing code, and that is a decidedly new departure for me, but again, something that is quite fun.
Another option that I'm looking at is including ChatGPT prompts in some posts here. In particular, I'm developing prompts that will turn the content of a blog post into a letter that you might then be able to send to your MP. All you would need to do is add your name and address, because MP will not respond unless you do so, plus the URL of the post, or better still the text of that post, copied and pasted from this blog, and then poist that information into ChatGPT, and it will produce a letter. Trials with this look to be very good. If this works, I will be writing other posts on this very soon.
Variations on the prompt would also produce letters to local newspapers, for example, or blog posts based on my own for use on other blogs.
Given that my goal is to get the ideas created here out into the wider world, these things seem to make sense to me whilst helping turn people from readers into active campaigners, which has to make sense. That is why I might also include prompts to create social media posts. None of these prompts will, of course, remove the need to both check and to refine the material created, but from what I've seen so far, the output from this process can be very good and worth using.
It is, however, by having thought through the entire process in advance, and having worked out what AI is really meant to do, which requires a prior thinking process, that it can really deliver.
Finally, and I am already doing this, I am using ChatGPT to assist research on blog writing. I like writing, but cannot see why asking AI for an opinion on the data to use in a post, and even how to structure it, is very much different from using Google these days, and I have used that resource for blog writing ever since I began writing here, 19 years ago. So, whilst I have absolutely no plan to use AI to write blog posts (although I have discovered that I can train it to write them in a style that is remarkably similar to my own, and which sounds like me) I will, undoubtedly, be using it as a focused research tool and I am planning to develop yet more AI prompts to achieve this goal.
Finally, if anyone is interested in how I might write those prompts, I might, in due course, make videos on that subject. These types of videos appear to be popular on YouTube at present, but they might become part of the member-only materials that we are thinking about publishing.
I hope that makes sense.
Comments are welcome.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
I hope you had a good holiday – and that the trains were not affected by fire bans
No trains this time….
Agree with all points. I use ChaGPT to develop talking points/analysis on electricity market reform. I have trained it. Now, I can get an analysis of turgid nonsense that, for example, the European Parliment issue (they are voting on a motion to do with power networks) . ChatGPT analysed the motion (based on discussions/training) provided a one pager – weak points good points (now circulated to the EP) and a one pager with proposals for change. Took about 15 mins. But it took hours/days to train the AI to this point.
Now when the European Commission (or others) generate the usual nonsense on market reform – I’ll speed read, mentally pick out some bits and pieces then pass it to the AI for analysis.
BUT.
I can only do this because I have a great deal of background data.
I am enjoying the training.
I am now using iterative prompts, i.e. promopts to write prompts that improve prompts. It works, but you end up wih prompts of 1,000 words…
Then you work out how to shorten them.
I confess to being really spooked by AI – a combination of Luddite prejudice, ethical concerns (energy), a well founded mistrust of those who own the organisations using it (interesting story today about a WhatsApp customer service bot handing out a real unrelated person’s private mobile number to someone asking for a train enquiries phone number – eventually the bot “confessed” to an ethical failure, although whether it repented too, remains unclear), and a suspicion of the manipulative power of algorithms, especially in relation to politics or marketing.
But as those concerns are all based on my almost complete ignorance of the subject, I trust you to make cautious use of it, because it certainly isn’t going away.
“ChatGPT ergo sum” as the great Zuckerberg never said.
Like all tools, AI has to be used with care.
Hamnmers are great, but not in the wrong hands.
throwing in the odd ‘Hamnmer’ tells us who wrote it 🙂
🙂
Your description of how AI is being (mis)used resonate with me. I have had dealings with two phone companies in connection with fibre optic installation at home, and with government and corporate websites in connection with my being executor for a recently deceased colleague, and I have been frustrated at (1) the vast number of web pages I have had to trawl through seeking answers, and (2) how hard it is to get to speak to a live human being when that search fails to provide the answers I need. It is clear that large organisations are using AI to generate responses that are irrelevant to the question I am trying to get answers to, and which do not seem to me to be particularly unusual questions.
I suspect that things will improve in time, as has happened with satnav systems, which in recent years have got much better both at recognising where I am and at instant rerouting of my journey when I divert (either intentionally or unintentionally) from their recommended route.
Hi Richard
Long time blog reader and follower here
Have you considered ways of enhancing the visual experience of your content – maybe AI in combination with other software packages can help?
They are many visual learners out there and understanding, for example government money creation, conciliation of govt accounts, role of bond issuance etc can be a challenge in just word form whereas as graphical representations can be easier to grasp for some
That is really hard to do….
It is on our agenda though
Can AI be used to emphasise the climate and other environmental implications of *every* policy (and lack of policy)? Especially looking ahead. Similar thought about inequality within nations and between nations.
Only if we tell it where to look, and define precisely what your terms mean. To presume it would find them and know what you mean is setting it up to fail. To get a good answer from AI requires careful definition of the question and data. It is not a miracle cure. If you can define those terms I can see what can be done.
One frustrating thing I have encountered with ChatGTP is after you have completed a search, you cannot copy and paste the search URL and forward it to share with colleagues because the URL is inoperable.
Any ideas?
I can….maybe you have to pay?
Can AI help with the complexity of all the methods of the advertising, packaging and marketing of food – and ultra processed food (UPF) in particular? It’s all made worse by the capture of regulatory bodies by industrial food producers who in turn may be owned by capitalists insisting on maximising profits to detriment of the nation’s health.
In what way help? Can you refine the question, Joe?
I claim my ten quid – it’s not Richard answering – it’s his A.I.
🙂
No it isn’t
Tyois will appear to prove it
I’ll have to give it some thought – and there’s a recently-planted wildflower meadow to be visited this morning.
However, I’ve used your template to write to my MP about Attenborough’s film ‘Ocean’ and Monbiot’s Guardian article about it.
To an excellent letter – created in seconds by Chat GPT – I have added:
On Thursday this week, about 30 of us met in West Bay to hear Doug Skinner (Purbeck resident and Greenpeace speaker) who delivered an informative and inspiring talk on the Health of Our Seas. He has offered his overview of ways to contribute “to a brighter, cleaner future for all”, Our Green Journey,
https://planetpurbeck.org/our-green-journey/
So, the MP will know, it’s not just me that cares about this.
I am delighted you have used it.
I am uncomfortable with your suggestion that it’s OK to use the eye-watering amounts of energy and water that AI consumes, as long as you compensate by other reductions. Climate action is no longer a matter of trade-offs. We should all be doing everything we reasonably can to reduce our carbon footprint.
You may reasonably argue that your work is important enough to justify the use of AI to help you spread your messages, but don’t pretend that you are not adding to the climate emergency to do so.
I am not doing so.
But you should not be using the web if you are concerned about electricity consumption. Nor should I be encouraging you to do so by writing this blog. Are you sting I should stop blogging?
A fair question, and one that worries me. I shall review my Internet use with the aim of minimising it.
Its a warm day, unfortunately this has closed the Rail lines around Cardiff. As my friend said in reporting this – What would teh Victorians make of us. Change, Change, Change is needed for the better.
Trains were a lot slower in Victorian times, and the odd derailemnt was toleated, In fact, many locos carried kit to allow derailements to be done without calling out a repair gang. That would not happen now.
Sure explore using AI, but be aware that at some point, for it to continue, the companies behind the various AI models will need to charge, and probably charge quite a bit, given the size of investments made to create the compute power needed to build the models, let along run the models afterwards.
What price would you be prepared to pay for the return that AI might give you?
For example, OpenAI (ChatGPT creators) raised $40bn earlier this year. Those investors will want a return on their money and sooner rather than later. OpenAI’s revenues last year was $4bn. How much longer can they offer ChatGPT for free?
I pay for ChatGPT
Serious users do
That is your answer
I realise you said that you couldn’t find the best video (with the French lady), but perhaps AI can help you find it?
Joking aside though, as you have clearly invested a lot of time in your research, perhaps a YouTube playlist or other form of “videography” would be helpful for others to allow them to watch the same ones?
Either without comment or with a bit of a summary-again, AI could do the summary…
Looking forward to the continuing high quality content.
That, I admit, is not how I intend to use AI.
https://northeastbylines.co.uk/business/technology/godfather-of-ai-now-fears-its-unsafe-he-has-a-plan-to-rein-it-in/
Interesting. Only AI can monitor ChatGPT successfully. But ChatGPT cannot play chess.
I read a couple of articles about AI data centres in the north east and Yorkshire using that much water that we will not be able to reach net zero.
Is that a problem for increasing AI usage?
That is a problem of energy usage generally.
And the issue is how we use heat generated. As Mike Parr has often pointed out, there are creative responses to that.
[…] I mentioned yesterday, I am making increased use of AI. One of my reasons for doing so is to increase the usefulness of […]
Hi Richard,
I think your idea of using your readers and viewers to spread your thoughts to MP’s and others is a good one. Make the best use of what talent and energy you (and Universal you), have to benefit others. Conventional media seems to be politically bias and not picking up ideas that challenge the status quo.
best wishes, Ann
Thanks
Lots of talk about AI at this conference yesterday.
https://www.mysociety.org/2025/06/19/tictec-2025-energy-open-exchanges-and-motivation-to-keep-pushing-forward-tech-for-democracy/
Do you follow mysociety?
I admit, I don’t.