folio

For all my critique of some LLMs and their implementation, what I do appreciate about the successful platforms is their providing the support I’ve lacked throughout my career.

Ever since my technical team was disbanded in 2011, I have been forced to go it alone. For every problem faced, I have had to work out a solution on my own, with no one to call on for help.

At times, those challenges seemed insurmountable. If I reached out for assistance, colleagues would always reply, “You’re the expert,” and leave me to it. Though I usually worked it out in the end, I often felt isolated dealing with immeasurable stress alone.

Enter competent LLMs, and I finally have a helpful colleague off whom I can bounce ideas. My very own sidekick that can now help me get back on track.

We may say that the years of going it alone were character-building,  helping me to grow independently and become more self-sufficient. Still, I would have much preferred a technical colleague when faced with all those challenges.

So thank goodness for the next best thing. The closest I am ever likely to come to a supportive team, it seems.

Honestly, Satya Nadella’s vision for the future of the web sounds awful.

I’ve just listened to an interview with the CEO of Microsoft, in which he opined enthusiastically about his vision of an “agentic” future.

A future where AI systems browse the web on our behalf, handle tasks for us, and reshape how we interact with information.

Coming shortly after an announcement that the company is laying off 6,000 people, I wonder how many share his optimistic vision of a future empowered by AI agents.

In truth, this vision of the web isn’t empowering at all. It’s reductive: another step away from the open, human-centred web many of us grew up with and still believe in.

That was a web where people share, explore, and connect directly. Not through corporate intermediaries, or through machines trained on our unconsented labour, rewriting our words into something marketable.

The web we knew was never meant to be a polished feed or an AI-curated experience. It was meant to be open, democratic and inclusive.

Yet now, big tech proposes putting a layer of AI between us and everything. Our voices, sites, and choices, filtered by systems most of us didn’t ask for, built on data we shared freely when we still believed the web was ours.

I’m not against innovation, but I do oppose a vision of the future that treats people as inefficient, unpredictable, and disposable, all while calling it progress.

The open web began with the promise of a shared, decentralised space where knowledge, creativity, and connection could thrive.

Over the years, we’ve watched it shrink. Social media turned the open web into walled gardens, locking users into closed platforms where algorithms fed on their content.

Now, with agentic AI, big tech is proposing to break it even further. Microsoft and others envision a future where users no longer explore the web for themselves.

Instead, these corporate “agents” will do it for them, reshaping the experience into something abstracted and opaque. A web filtered entirely through the lens of big tech.

For many of us — users, writers, site owners, and advocates for inclusivity and accessibility — it sounds dystopian, severing people from the diversity and richness of the human web.

Worse, it centralises power even further, handing the keys to AI companies whose models were trained, without consent, on the very web they now seek to control.

An open web is a democratic ideal: messy, human, multilingual, and full of surprises. It should be protected, not just for developers or content creators, but for everyone.

When access, agency, and diversity are lost, so is the soul of the web. What could be worse than a future where corporations and their agents speak for, to, and over us?

Naturally, many of my colleagues do not agree with my assessment of Microsoft 365 Copilot.

Unlike me, they find it extremely useful, saving them hours each week to focus on their core responsibilities.

Which is wonderful. I am all for tools that make people’s lives easier. We all have different skills and needs, after all.

Still, given the current capacity of LLMs to hallucinate, be that incorrectly summarising information or fabricating citations, I remain wary of some of their use cases.

Though early adoptors rarely want to hear about regulation, there’s good reason to slow down and await assurances that the technology is safe.

Teams that have established new workflows underpinned by AI feel they have been thrown into disarray by a new, more cautious approach.

Fair enough. But in any other area of service change, the organisation would first have established mechanisms for validating and monitoring such tools, long before putting them to use.

Copilot is sold as a productivity tool to aid existing users of the Microsoft 365 office suite. Some people find that to be the case. They find its ability to summarise and draft content invaluable.

Others cite the benefit of tools such as live transcription, forgetting this was a feature already baked into the platform long before Generative AI was ever on our horizon.

In truth, Copilot is largely just a new user interface and integration layer — poorly implemented — built on top of existing tools, primarily to placate shareholders rather than actual users.

I’m glad that some of my colleagues find it useful. Some value for money then. But I must say I wouldn’t be surprised if the whole platform went the way of Cortana before it. To me, it’s already a dead donkey.

In the 1990s, my family was early to embrace the concept of Fair Trade long before it was fashionable to do so.

I wonder if my generation needs a similar movement to counter the unethical behaviour of big tech.

Despite widespread acclaim, there’s a lot wrong with the leaders in AI in their current form.

Their models were trained on vast quantities of stolen data — text, images, video and audio — in clear breach of international copyright law.

And worse: they operate digital sweatshops in developing nations exploiting cheap labour to sift through, categorise, and filter the data inputs and outputs of generative AI.

Most people, of course, have no idea about the human cost of implementing these technologies, much as people used to be ignorant of the wages earned by producers of our food and clothing.

Sold on the hype, most of us believe the technology itself is squarely in control of its own moderation processes. Most would be shocked to learn that behind the scenes, it is highly reliant on exploitative labour practices.

Fair Trade began during a collapse of global coffee prices. Thirty years on, what will incentivise us to demand ethical tech? Only a collapse in our own labour market?

I’m not opposed to the use of Generative AI. I think it has great potential as a productivity aid and assistive tool.

Platforms like ChatGPT and Claude have shown real promise in making certain tasks more efficient, insightful and accessible.

But I remain reticent about its global adoption. More than anything, I object to it creeping into every aspect of digital life.

I don’t want AI integrated into my operating system. I don’t want it embedded in my word processor, my web browser, or my website’s backend.

I certainly don’t want a flood of AI-generated content dominating platforms like YouTube, LinkedIn, or Medium — although that is already happening.

Automation has its place, especially in the workplace, where it can streamline tasks and improve productivity.

But in my personal life, I want space to think and reflect. To learn slowly, and explore tentatively.

I want to create rather than consume, and splurge my thoughts without perfection. The way I always have.

It seems that venture capital and big tech are racing to insert AI into every interaction, screen, and task, whether we want it or not.

But, in truth, not everything needs to be optimised, monetised or enhanced. Sometimes, we just want to be left alone to live our lives independently.

Give me platforms and tools free of synthetic content. Offer me tools that do not nag at me to delegate my intellectual endeavours to an algorithm.

Understand that not everything has to be frictionless. Ask why AI is being added everywhere, not just whether it can be.

This is neither nostalgia nor fear. It’s a question of balance. I am open to a future where AI is available, but not one where it is unavoidable.

I want to retain my sense of self, and be able to think for myself. Most of all, I just want the tools to disappear into background, like the best unobtrusive technology.

In short: give me space to breathe.

The first time I encountered the term “Artificial Intelligence” outside the realm of science fiction was in 2003.

That was when I was introduced to a family friend who was doing a PhD in the field, then working with a medical imaging company that was developing algorithms that could detect cancer tumors in CT scans.

He had a proposition for me: he and a fellow PhD student were setting up an independent IT business and wondered if I would like to join them.

This was more a charity to me than an acknowlegdment of any tangible skills I had, noting that soon after marrying two years earlier, I had been made redundant and had oddjobbed ever since.

Though wondering what benefit I might bring to their learned company, I took them up on the offer, noting my better half’s immeasurable patience with the inconsistent income I brought into the household.

Only, it very quickly materialised that there was a clear disagreement about direction of travel between the two business partners.

The partner endowing the bulk of the funding — still gainfully employed by the medical imaging company — had visions of a software startup offering IT consultancy across London.

His partner, striking out on his own meanwhile, was intent on converting his fledgling internet café into a coffee shop. And that, it turned out, was where I came in.

In the end, I was not travelling early morning across London from West Ealing to Harringay to put my Masters in Publishing to good use, but to drive an Italian coffee machine.

And so it went on for months. The closest I got to touching any IT was manning a terminal for a bank of computers from my perch opposite The Salisbury Hotel on Green Lanes.

Eventually, our family friend, growing increasingly frustrated with his business partner’s hospitality focussed venture, pulled the plug on funding the business and advised me to do the same.

Cutting his losses, he went back to focussing on his work developing algorithms to detect tumors in CT scans, now a very lucrative field of research.

And me? I did a stint working in a warehouse where nobody got paid, another managing a café in Mayfair, and finally got a job in the NHS sticking labels on envelopes. Yes, my twenties were a great success!

Perhaps these experiences explain why I have become attuned to spotting false promises. A worthy talent to have.

Now more than ever, in this age of the big tech and venture capital AI bubble, it’s important to have fully functioning bullshit-detection capabilities.

Both big tech and venture capital have sold the world false promises the past two years. Some are calling it the biggest scam in history. Others liken it to the dotcom bubble of the early 2000s.

Either way, a crash is on the horizon. Why? Because it has simply been oversold. Some of the tech is cool and useful, but much of it is the complete opposite.

It’s like the difference between a CT scanner and an Italian coffee machine.

Everybody is belatedly realising that AI chatbots are just super impressive demos. But that’s it.

They’re not actually ready for real-world use cases, because they’re neither reliable, nor consistently produce accurate outputs.

I wonder if that’s why Microsoft recently cancelled it’s vast data centre expansion plans. And why Apple fired its AI development leads.

LLMs are incredible, without a doubt, but it turns out they are not very good at performing the kinds of tasks they’re being marketed for.

A tough realisation for those who have invested everything in this bubble. A bit of a relief, though, for those companies that are pulling out because it’s just too expensive.

The hope is that the insane hype is abating at last as the novelty wears off. Yes, it was a great demo for a time. But let’s all now snap back to reality.