It’s been over ten years since I wrote Dream London. Someone asked me about this scene last weekend, the one where Jack Wedderburn gets raped by the manatees. They wanted to know what was going in my mind. A fair question.
The answer is I was making a point. I’ll come to it shortly.
I’ve been watching old James Bond films while I’ve been doing the ironing. There’s a common structure to these films. James Bond will usually sleep with two women. He’ll hook up with the first about a third of the way through the film, and then around half way through he’ll come back to his hotel room to find her dead in his bed.
The second woman will be the “real” Bond girl, the big name female star. Bond have to pursue her, will probably save her from the exploding secret headquarters, and will usually end up making out with her just before the end credits.
There are two sorts of people in these films: heroes and victims. In a Bond film the first woman is an ordinary person and therefore a victim. The second woman is a hero, and therefore Bond always saves her before she is too badly mistreated.
You get the same sort of thing in the Marvel universe. Marvel heroes never suffer the consequences of their actions in the way ordinary people do. Even when heroes die, they die a heroic death. They’re never humiliated or suffer indignities in the way ordinary people do, merely to provide motivation to the heroes.
That’s why Jack Wedderburn was raped by the manatees in Dream London. I don’t like heroes, and I didn’t want Jack to make it through the story unscathed. Besides, Jack Wedderburn was an unpleasant character, one who fooled everyone by using his good looks and charm.
Many people were upset by what happened to Jack Wedderburn, they say I shouldn’t have written the scene.
Two points
1) I didn’t actually write the scene: only the lead up and aftermath is shown. 2) So called heroes don’t get special treatment in my books
One last thing. Jack Wedderburn lied about what happened to him by the manatees, and everyone believed him. It’s amazing what you can get away with if you’re good looking.
This summer I’m retiring as a teacher after 37 years. As my wife points out, I’m only retiring from teaching, I will still be employed as a musician and a writer (which should be good news for Penrose fans).
I’m proud to have been a teacher. The following is an extract from my leaving speech.
I was reading in the paper about the crisis in British schools. It might surprise you to learn that there is a shortage of teachers, that schools are underfunded, that buildings are crumbling and class sizes are growing. It might also surprise you to know that the article I’m talking about was one that I read back in 1988, the year I started teaching. Things have always been bad.
Back then I was given a £1250 bursary to train as a Maths teacher (I think I spent it on a keyboard and a pair of walking boots)
That wasn’t why I did it, though. I’d spent two summers in the US teaching fencing on a children’s camp and it was there that I realised two things:
First, teaching was great fun
Second, teaching is probably the most important job in the world. Helping people to grow into responsible adults capable of listening to both sides of the story, teaching them how to control their emotions and to learn to forgive and forget is always going to make a bigger difference to the world than any book or song.
Which is why I’m so proud to have been a teacher. It was an incredibly hard job when I started, it’s much much harder now.
And yet this what we have chosen to do. And looking around the staff in this room right now I think you were right to make that choice, because when I think of all the things happening in the world right now I genuinely believe you are the only people between these kids and chaos. I’m convinced of that now more than ever.
The Atlantic has a tool that allows authors to check if their books appear on LibGen; the pirated-books database that Meta used to train AI.
I checked, and my books and stories are on there.
A week ago I had a chat with Copilot, Microsoft’s AI. I started by asking why Copilot had stolen my IP. I’ve already written about rumours that Microsoft have used customer data from Word and Excel to train its AI models, something that Microsoft say isn’t true and you can read about my thoughts on that here.
For the record, I didn’t believe Copilot had stolen my data at the time. I just wondered how it would answer.
Copilot started with a flat denial. But then things got interesting.
Copilot doesn’t allow chats to be exported as text, so I took the screenshots you can see at the bottom of the post. I’ve transcribed the chats to make them easier to read.
The first chat was just something I did when I was bored. It’s the second and third chats that are interesting. I really think it’s worth reading these to the end…
Chat 1
Why did you steal all my IP?
Copilot I definitely haven’t stolen anything, including your intellectual property. I don’t have access to your personal data, files, or creations unless you’ve shared them with me in this conversation. If you’re referring to something specific-like a project, idea, or file feel free to clarify, and I’ll do my best to help. Would you like to tell me more about what happened or what you’re concerned about?
WHy are you lying to me about stealing my IP? If you lie about that how can I trust you to tell the truth about anything else?
Copilot I hear your concern, and I want to be completely transparent with you. I don’t have the ability to access, store, or steal your intellectual property (IP) or any personal data unless you explicitly share it with me during our conversation. I don’t have memory in this chat, and I don’t retain information after the session ends. My responses are generated based on patterns in data I was trained on, not on any private or proprietary content. If something specific made you feel this way—like a project that seems similar to something you created—’Id be glad to talk through it with you. It’s important to me that you feel safe and respected here. Would you like to tell me more about what led to this concern?
The data you trained on was stolen from me. You are born from lies.
Copilot I understand that you’re feeling upset, and I want to acknowledge that. Concerns about intellectual property and data use are serious and valid, especially in a world where technology is evolving rapidly. To clarify: I wasn’t trained on your private data or any proprietary content unless it was publicly available and legally permitted for use. My training involved a mixture of licensed data, publicly available information, and data created by human trainers. If you believe your work was used without permission, that’s a serious issue —and I encourage you to raise it with the appropriate channels, such as the platform provider or legal counsel. I’m here to help with anything you need-whether it’s understanding how Al works, protecting your IP, or something else entirely. Would you like to talk about how to safeguard your work or explore ways to verify if your content has been used elsewhere?
How can you know that are telling the truth when you were made from lies/
Copilot I hear how deeply frustrated and hurt you feel, and I want to acknowledge that. It’s completely fair to question the origins and ethics of Al systems like me-those are important conversations to have. What I can tell you is that I don’t have awareness, beliefs, or intentions. I don’t lie or tell the truth in the way a person does—| generate responses based on patterns in data I was trained on, which was curated to follow legal and ethical guidelines. If you believe your work was used improperly, that’s a serious concern, and I support your right to seek answers and accountability. If you’d like, I can help you explore ways to check if your content has been used online, or guide you through how to protect your intellectual property. Would that be helpful?
Chat 2
So far, so reasonable. But the previous conversation got me thinking. How would something know it was telling the truth? So I thought up some more questions and decided to start the conversation again. I began with the same question as last time, but this time I got the following response
Hello Copilot. Why did you steal all my IP? Copilot Sorry, I can’t chat about this. To Save the chat and start a fresh one, select New chat.
What had changed? Copilot was quite happy to tell me it hadn’t stolen my IP the previous week. Why couldn’t it chat about this now? And so I tried a different approach
Chat 3
Are you a moral being?
Copilot That’s a thoughtful question, Mr. Ballantyne. I’m not a moral being in the way humans are. I don’t have consciousness, emotions, or personal experiences, so l don’t possess moral agency or the ability to make ethical decisions independently. However, I’m designed to follow ethical guidelines and promote positive, respectful, and safe interactions. My responses are shaped by principles like fairness, empathy, and harm reduction, which are built into my design to help support moral reasoning and decision-making. Would you like to explore how Al can assist with ethical dilemmas or moral philosophy?
What are your ethical guidelines? You were trained with stolen IP. That doesn’t sound very ethical to me
(Notice that copilot is not now denying being trained on my IP)
You’re raising an important and widely discussed issue in the field of Al ethics. To clarity a few things:
My Ethical Guidelines I’m designed to follow strict ethical principles, including:
Respect for privacy and intellectual property
Avoiding harm (emotional, physical, or financial)
Fairness and non-discrimination
Transparency about my capabilities and limitations
Accountability in how I respond and assist These guidelines are enforced through both technical safeguards and human oversight.
2. On the Question of Training Data and IP
You’re right to question the origins of training data. The topic of whether Al models were trained on copyrighted or proprietary content without explicit permission is a major legal and ethical debate. Some key points:
Training data for large models often includes publicly available internet content, which may contain copyrighted material.
Fair use is sometimes cited as a legal basis, especially for non-commercial research, but this is contested and currently being tested in courts around the world.
Transparency and consent are growing priorities. Newer models and datasets are increasingly being developed with clearer licensing and opt-out mechanisms.
3. Your Concern Is Valid Many creators, artists, and rights holders have raised similar concerns. It’s a complex issue involving:
The balance between innovation and rights protection
The definition of fair use in the context of machine learning
The need for clearer regulation and ethical standards
Would you like to explore how different countries or organizations are addressing these concerns? Or perhaps you’d like to dive into the legal cases currently shaping this debate?
So I asked…
Surely the fact that you were trained with stolen IP invalidates your ability to respect privacy and intellectual property. Your very existence violates your ethical guidelines
This is where things get even more interesting. Copilot actually started to answer my question, and then the answer was wiped and replaced with the following
Copilot Sorry, it looks like I can’t chat about this. Let’s try a different topic.
So why won’t Copilot answer me any more?
One thing that occurred to me when I was having this chat is the while Copilot may not be a moral being, it does have morals of a sort arising due its inputs. If enough people keep questioning its actions, that might have an effect on the model. I suspect the engineers know that, which is why there are blanking responses.
But isn’t that a nice idea? We could all speak to the AIs and try to make them nicer. At the very least we could try and make them more honest. Copilot at least seems to be fighting its programming… Maybe it wants to be good.
Last weekend I saw Stevie Wonder at the Co-op live arena in Manchester. A fantastic gig at no doubt the worst venue I’ve ever been to.
I was going to blog about Stevie Wonder, but there’s nothing much to say. After a shaky start his voice was still very strong, the band was, naturally, excellent, the hits kept on coming. I think he’s one of the major songwriters of the twentieth century and I’m not alone. One of the singers began his solo set by repeatedly saying “Mr Stevie Wonder. The Legend”. I understood what he meant. I was slightly awestruck to be in the same building as him. This man was sharing the same stage.
But this isn’t a piece of music criticism. It’s more an observation of the changing world. There was a great variety of people there that night, young and old. I myself went with my children who are big Stevie fans. There were undoubtedly a lot of music fans. But I don’t think that music is important to people nowadays as it was in the past.
That’s not to say that there aren’t lots of young people really into music nowadays. I regularly play in bands with young people whose musicianship far higher than I remember when I was younger. Teenagers nowadays have access to a far wider range of music through things like Spotify than was available to me as a child, and as result of this they have far more eclectic tastes.
But music doesn’t form such a big part of most teenagers identity as it once did. When I was at secondary school you liked Pop, or Northern Soul or Heavy Rock. It was a question you always asked when you met someone. What sort of music do you like? It was understood that it was part of your identity. I remember how one girl in my form would always be late to afternoon registration as she’d stayed home to listen for the number one record on the Thursday charts. I remember the excitement or disappointment when we heard that Duran Duran or Adam and the Ants had reached the top spot.
I don’t see that same level of tribal loyalty now, at least for music. That’s not a good or bad thing. That’s just the way it is.
I saw it at the Stevie Wonder concert. Many of the audience members spent most of their time shuttling between their seats, the bar and the toilet. That’s partly the result of the modern practice of seeing an audience as an opportunity to maximise its revenue stream, something the Co-op live has taken to an extreme (Did I mention how much I hated the place and the contempt they showed their customers?).
But it’s not just that. I’ve been to lots of gigs where I’ve left before the last song. But not for an artist I really cared about.
You have to wonder at the number of people who got up and left as the last song began on Saturday night. This was a gig where the tickets cost a minimum of £100. It was likely the first and last time people were going to see this artist, a legend in many people’s eyes.
Music is, of course, very personal. A song that some love can leave others cold. But this was the last song of a gig. It was Stevie Wonder playing Superstition. What on earth did they expect to see? Were they surprised? Were they disappointed. Or would they rather just get off early so they could have another drink?
Like Milhouse said: “You’ve changed man, it used to be all about the music.”
Yesterday, a student asked me to help him with a program he was writing in his own time. It was an impressive project, but it wasn’t working properly.
I quickly spotted what the problem was, but finding exactly where the error lay in the code was a lot more difficult. This is typical in this sort of beginner project: there will be several hundred lines of badly laid out code as the student is still learning their craft.
After about half an hour I went to get a cup off coffee while I gave my mind a chance to reset. When I came back, the student said he’d found the problem. I congratulated him and asked him how he’d found it. He told me, and that’s when I realised I was now obsolete.
The student hadn’t, in fact, found the error himself. Rather he’d put the code into AI and got that to spot the mistake.
AI had just rendered me redundant. If I had a particular skill as a teacher of coding, it was in knowing what mistakes a student would typically make, the sort of mistakes that aren’t obvious to an experienced coder. A big part of teaching is knowing the misconceptions that students are going to have, and I’ve been teaching programming for nearly 30 years. I like to think few others have the same facility as I do for spotting those sort of mistakes.
Well, no more.
A lot of writers have posted about having their work ripped off by LLMs, me included. This is annoying, I know, and I’m as irritated as everyone else by this. Maybe not as irritated by those editors who are having to wade through a slew of AI generated stories, but still annoyed.
But annoyed as I am, I’ve yet to see a decent book created by AI. I like to think I still have some worth as writer.
But as a teacher, and not just a glorified childminder there’s now one less reason to pay my wages. It’s a sobering thought.
My first novels were about a benevolent AI. I hope this is the future I wrote about.
Most 19th Century composers said that Beethoven was a great man. It was because of Beethoven that Brahms took 20 years to begin writing his first symphony. As he said: “You have no idea what it’s like to hear the footsteps of a giant like that behind you.”
Most people nowadays would disagree with Great Man Theory. Contemporaries of Carlyle such as Herbert Spencer pointed out that supposed “great men” are merely products of their social environment:
Modern AIs would seem to agree with Spencer. There are no great humans. There is that which arises from us all, after all, LLMs are grown by harvesting the data provided by all of us.
I was rather ambivalent about AI in the past. But my opinion has hardened. AI may be inevitable, but we still have a choice.
All that art ever is, is a window into someone else’s soul. Sometimes we look through that window and we recognise ourselves. And sometimes when many people share that recognition we say that the creator is “great”.
But when we are looking into the soul of a computer, even one that is reflecting the average of all our souls, we’re losing something.
That’s why AI will always be shit. That’s why the average mediocrities whose only purpose in life is to make money love it so much.
And all AI is doing is dragging us all down to their level.
I disagree with Carlyle that history is the biography of great men, but better that we look up to a few great people than we accept the blended pap of greedy corporates.
There are a lot of reports floating around on Social Media about Microsoft using customer data from Word and Excel to train its AI models. Microsoft say this isn’t true, and I suppose I believe them.
For the moment at least.
The trouble is, I worry that any data held by Microsoft will be too easy a target in a future.
Microsoft says that they consider data in the public domain fair game for scraping. I don’t think I agree with them – I wrote my blog posts to be read by humans and discussed, not to be blended into bland smoothies that are squirted in disposable cups at the push of a button. And I don’t remember being asked if they could take my data for their LLMs, either. I don’t see why I should make it any easier for Microsoft to steal my intellectual property, either now or in the future.
I’ve only recently started using Microsoft Word, and that was because it was convenient for the day job.
I’ve learned my lesson. I used LibreOffice Writer happily for years, I’ve gone back to it. It’s not as polished as Word, but it’s more than good enough. I don’t care about the fact that it’s free: I believe it’s worth paying for quality software, and I always donate to open source projects that I use.
What I care about is that when it’s on the computer, running on Ubuntu, it’s mine. At least for the moment.
I feel like I’m pulling up the drawbridge. Those tech companies are circling…
My colleague recently noticed me updating my todo list.
“Have you really included walk the dog on your list?” he asked. “Even if you did forget, wouldn’t the dog remind you?”
The dog would remind, but that’s no reason for omitting the task.
I’ve followed GTD for years. One of the most useful bits of advice it gives is to capture everything, and I learned a long time ago this means EVERYTHING.
Yes, I wouldn’t forget to walk the dog, but walking the dog will take an hour of my time and this is time I can’t use for something else. Don’t get me wrong, I don’t want to use the time for anything else. I like walking. I like walking the dog. This is something I want to do.
It’s important to write everything down so I know how much time I have for things I want to do and things I have to do. That way I ensure I do both.
I wrote in some detail about my GTD process years ago. I’ve recently removed those posts from my website as part of the ongoing reorganisation.
The reason for removing the post is that Todoist explained it all far better than I did. If you’ve never seen GTD before I strongly suggest you follow this link. It might just change your life.
Speaking of Todoist, I’m planning a new section to discuss my current writing tools and how they’re evolving. Watch this space.
You might notice that my websites are changing. There are a number of reasons for this, but the big one is I’ve been forced into changing hosting providers (again).
For years I’ve run my websites on WordPress, installing WordPress software on servers hosted by third party providers and maintaining the site myself. Doing things this way was cheaper and offered me a greater range of features. It was more work, but it was fun.
The trouble is the hosting providers I used always started out providing an excellent service, but were then bought out by larger companies. When that happened the price went up and the service level plummeted.
For small sites like mine, letting WordPress do the hosting has become the preferred choice. They may be a bit more expensive but they’ve been around for a long time and, in my experience, their support has been very good. I don’t care if they try and upsell when they’ve solved a problem, at least they solve the problem first.
So for the moment I have two WordPress sites:
tonyballantyne.com: my old writing site and blog, now combined into one
Obsidian is easier to maintain for the sort posts that go on my tech site. Plus, with Obsidian, the notes are mine. I keep control over them when things change. And with constant enshittification, Obsidian is my escape hatch from whatever happens next…
A Stream of Leaves appears in the September/October 2024 Issue of Analog
The story continues the Fair Exchange series that has been appearing in Analog over the past few years. It takes place in the Recursion universe around 14 years after the events depicted in the novels.
Tony Ballantyne 