A long time ago, I cancelled my social-media accounts. Not merely one troublesome platform, but all of them. LinkedIn survived longer than the others because it still appeared to have some professional purpose. Recently, however, I cancelled that as well.
I also went back through what remained online and deleted every old post I could find that no longer seemed relevant. Years of fragments, announcements, reactions and observations disappeared. I did not feel that I was destroying an important archive. In most cases, I felt that I was clearing away material that had already become functionally invisible.
I now use my blog.
That may sound like a return to an older Internet, and in one sense it is. Blogs existed long before the modern social-media feed took control of how most people publish and consume information. However, the way I am using my blog is not nostalgic at all. It combines the freedom of long-form independent publishing with something very new: an iterative writing process built around artificial intelligence.
This article itself is an example. It began as a spoken thought on my phone. I described the idea, stopped, added more, corrected the direction and continued. There was even a small interruption when moving the phone sideways triggered an unexpected interface response. I laughed, resumed the conversation and kept developing the post.
I was not sitting in front of a computer trying to summon a finished article from an empty screen. I was thinking aloud while the idea was alive.
The freedom I had forgotten
A blog gives me room to think.
I can write a short update when that is all a subject requires, but I can also publish 1,500 words, 5,000 words or an entire series. I can use headings, colours, photographs, diagrams, quotations, tables, links, video and other media. I can return to an old subject years later and connect it to a newer one. I can organise posts into themes and preserve a chronological account of my work, travels, family history, technical projects and life in different countries.
Most importantly, the article is the destination. It is not merely an object inserted into a feed and pushed downwards by everything published after it.
Social media trained us to treat expression as a sequence of disposable moments. A post appears, receives some reactions and then vanishes beneath newer material. Even when it technically remains accessible, it often becomes difficult to locate, organise or place into a meaningful body of work.
A blog behaves differently. It can become a personal archive rather than a stream of interruptions.
I should add one qualification. Blogger is itself a hosted service, so using it does not provide absolute technical independence. A properly maintained blog still needs exports, backups and, ideally, a domain and content strategy that make future migration possible. Nevertheless, the format gives me substantially more control over presentation, length, organisation and reuse than a typical social-media post.
What I mean when I say social media is a lie
“Social media is a lie” is deliberately blunt, but I do not mean that every person using it is dishonest or that every interaction is worthless. I have met people, discovered information and found professional opportunities through such platforms.
The lie is the impression that I am publishing on my own terms.
I am not. I am supplying content to a system whose layout, audience, priorities and distribution mechanisms belong to somebody else. The platform decides how much text is comfortable, which formats receive attention, what appears in a reader’s feed and how long it remains visible. It can change those rules, introduce restrictions, remove features or discontinue a service.
LinkedIn’s own terms are quite explicit about this distinction. Users retain ownership of their original content, but grant the platform a broad licence to process and distribute it. The service also uses data to order recommendations and organise the feed, and it does not present itself as a permanent storage service.
That is a legitimate commercial arrangement, but it is not the same as maintaining my own body of work.
Social media also encourages performance. A thought is shaped not only by what I want to say, but by what might provoke an immediate reaction. Complex subjects are compressed. Qualifications are removed. Conclusions appear before the reasoning that supports them. Professional platforms can become theatres in which everyone is successful, inspired, humbled and delighted to announce something.
Real life is rarely so tidy.
My projects contain mistakes, uncertainty, bureaucracy, incomplete evidence, changed opinions and things that did not work. A blog gives me enough space to preserve those imperfections. They are often the most valuable part of the story.
AI changed the moment at which writing can begin
The most important change is not that ChatGPT can generate paragraphs. The important change is that I no longer need to begin with paragraphs.
I can begin with a complaint, a memory, an observation or an untidy collection of ideas. I can speak while walking. I can record a thought when I have stopped during a journey. When using voice in a car, this must of course be hands-free, permitted by law and safe under the actual driving conditions; the purpose is to reduce distraction, not create another one.
I can stop halfway through an idea and return later. I can add an example the next morning, correct a date, reject an interpretation or ask for an entire section to be reorganised.
The process is roughly this:
Spoken idea
↓
Additional memories and context
↓
First structured draft
↓
My corrections and objections
↓
Research and factual checks
↓
Revised article
↓
My final validation
↓
Blogger-ready HTML
The loop matters more than the first draft.
I do not give the AI a sentence and blindly publish whatever returns. I provide the experience, facts, intended meaning, arguments and conclusions. ChatGPT turns that material into a structured draft. I read it, identify what is wrong or missing, and send it back for another revision. Sometimes a phrase does not sound like me. Sometimes the structure is too generic. Sometimes an important qualification has disappeared. Sometimes the AI gives too much importance to a secondary point.
I correct it, and the article improves.
Over time, the system has accumulated extensive instructions about how I prefer to write: British English, substantial articles, clear distinctions between facts and opinions, protection of private information, careful treatment of disputes, useful research and a consistent Blogger-compatible HTML structure. Previous conversations and posts also provide continuity.
This does not mean that an AI magically understands me as another human being would. It means that I have created a detailed editorial context and refined it through repeated feedback. The quality comes partly from the model, but also from the accumulated discipline of the process.
The blank page is no longer the starting point
Traditional writing often begins with a physically and mentally demanding ritual: sit at a desk, open a document, remember everything that seemed important earlier and somehow arrange it into coherent prose.
That model has always excluded useful material. Ideas arrive at inconvenient times. Memories surface while travelling, walking through a familiar place or discussing something unrelated. By the time I sit in front of a computer, the detail or emotional precision may have disappeared.
Voice capture changes this. The first version can be conversational, repetitive and grammatically chaotic. It does not need to be publishable. Its job is to preserve the thought.
AI then helps convert that raw material into structure. It can identify themes, remove repetition, suggest an order and expose gaps. I can ask it to challenge a claim, verify a public fact, improve a technical distinction or make the conclusion better reflect what I actually believe.
Finally, it produces the complete article in HTML. I copy that HTML into Blogger, preview it, make any final adjustments and publish.
The transition from thought to publication becomes less dependent on a single long session at a desk. Writing becomes continuous and asynchronous.
The author still has the difficult job
None of this removes responsibility.
AI can misunderstand a dictated sentence, especially when speech overlaps with noise or when a conversation moves quickly. It can produce a convincing connection that I never intended. It can confidently introduce a detail that was not present in my account. When research is involved, a source may be outdated, misread or insufficiently authoritative.
There is also an aesthetic risk. If people provide only a vague instruction and accept the first output, the Internet will fill with smooth, competent and entirely forgettable prose. The text may be grammatically correct but contain no experience, tension, uncertainty or individual judgement.
The bottleneck therefore moves. Typing becomes less important; selection and judgement become more important.
I still have to decide:
- whether the idea deserves to be published;
- which memories and facts are relevant;
- what is observation and what is interpretation;
- which claims require verification;
- whether the draft genuinely represents my view;
- what should remain private; and
- when the article is finally honest and complete enough to publish.
What this may do to Internet publishing
I suspect this workflow will change content production far more than simply allowing companies to generate larger quantities of marketing text.
It reduces the distance between thinking and writing. People who have valuable experience but dislike typing may be able to publish substantial accounts. People writing in a second or third language can concentrate on meaning while using AI to improve presentation. Technical specialists can supply expertise without spending hours fighting with structure. Personal histories that would once have remained as scattered memories may become coherent records.
It may also revive long-form personal publishing. Social media won partly because it made posting effortless. Blogs offered more control, but required more work. AI changes that balance. A blog can retain its depth and flexibility while gaining some of the immediacy that made social platforms attractive.
The negative possibility is obvious: an enormous increase in generic material produced merely because production has become cheap. Search results, company sites and blogs may become saturated with articles that have no identifiable authorial experience behind them.
The positive possibility is more interesting. AI can allow people to document lives, projects and ideas that would otherwise be lost—not by inventing their stories, but by helping them articulate and organise what they already know.
That is what I want from it.
I am not trying to manufacture a public persona or feed a platform every day. I am building a durable record: of work, technology, family research, places, mistakes, plans, bureaucracy, renovation, travel and whatever else becomes important enough to preserve.
I left social media because it increasingly felt like producing fragments for somebody else’s machine. I returned to blogging because I wanted space, continuity and control.
AI has made that return practical. It lets me collect ideas when they occur, develop them through conversation, challenge the result and eventually transform them into a finished article.
Including this one.
The article itself is also an example of the iterative process it describes: I supplied the ideas by voice, reviewed the draft, corrected its direction and retained final editorial control. The selection, interpretation, opinions and conclusions remain my own.
Comments
Post a Comment