The art of winging it in the age of AI

It’s half five in the morning and my head is pounding. My stomach practices gymnastics as I reach for the Alka Selzer. My mouth is a culinary fusion of embittered disappointment and a skunk’s jock strap.

Despite having the run of my career, swinging for the fences and hitting every home run I could, I still found myself on the scrap heap.

My ASD was incensed at the injustice. My ADHD was wailing in the corner. The me in the middle was overwhelmingly sad.

I needed a new path forward.

My phone buzzes, and I see that my final payment has landed. I was richer than I’d ever been. Coming from poverty, you’d expect this to be a moment to celebrate. Instead, I knew every penny was earmarked for survival.

The next email was from Companies House. Redacted parts of my memory began to declassify. Following an outpouring of rage and grief to Claude, it’d bet me that I wouldn’t start a company. I’m still waiting for it to pay up.

I named it after my ancestor, Thomas Bayes. It seemed apt, not just as a homage to Thomas’ godfather role in AI’s evolution, but because I had no good goddamn idea what I was doing. To find the path, I’d have to take a page from his book and apply some Bayesian reasoning to the problem.

Three months and 200 rejection letters later, and the Bayesian take would be that no one’s going to offer me a job any time soon. Bayes, however, has cleared its first invoice (a mighty £300), netted me a box of wine, and BT has one paying subscriber on Substack.

The path I would end up walking, it would seem, is a rollercoaster.

Everyone is winging it all of the time

Leading on AI, you’ve got to be comfortable laying the track while driving the train. There’s no playbook here, no beaten path. Fortunately, this is nothing new to me.

During the bygone era of the Coalition government, my old boss and mentor James Mawson had entrusted me with the keys to Global University Venturing, the sister paper to his Global Corporate Venturing magazine.

The brief: investors wanted to know about university innovation, but no one knew what was going on inside the ivory towers. My job was to find out.

A month in and riddled with imposter syndrome, I was on a train to London to an innovation conference. I was absolutely bricking it. Doomscrolling, I chanced across a piece in the Guardian. David Cameron’s series of blunders should be excused, it argued, as research shows that everyone is winging it all of the time.

While clearly intended as a commentary skewering Cameron, I actually found it inspirational (the only time you’ll hear me say that about a Tory PM). It was still bouncing around in my head when I met Tom Hockaday, the former MD of Oxford University Innovation, while networking.

After some pleasantries and sharing our backstories, I worked up the nerve to ask him: “Tom, are you winging it?”

“You’ve rumbled me,” he said with a wink.

Over wine, he explained just how crucial winging it is to innovation. We drew in others – investors, academics, entrepreneurs – and posed the same question.

It was an interaction which changed the trajectory of my career.

It didn’t matter that I didn’t have decades of experience.

What mattered was turning up, giving it a go, and learning from my mistakes.

Or, as Tom put it, you’ve got to “hang up your ego with your hat in the morning”.

The wisdom Tom shared with me is something I’ve kept coming back to in both my AI and my entrepreneurial journeys.

The fear and anxiety of it all can be crippling. It’s not just my own path which is unclear, but the route we’re all on. Linear thinking is of no use in navigating our way out of the fog of uncertainty.

Instead, we must continue to put one foot in front of the other – to continue sailing the ship even though we have no map. The course through is forged from experimentation and the willingness to accept the pain as failure as a guide.

In short, I’ve gone back to winging it.

Just like David Cameron.

Only with less pigs.

Everyone has a plan until they are punched in the face

An oasis of calm in the maelstrom of 2026 for me has been Retro Rewind, a video game where you run a video rental store in the 90s.

You scan returns and rewind tapes. You recommend classics to customers. Your pristine new releases display up front belies an unspeakable abyss you call a filing system in the back room.

I ran such a veritable enterprise in my late teens. In my first two weeks, the rest of the staff went for drinks one night. Long simmering resentments came to the fore, and the ensuing implosion let to me receiving a field promotion from part time to store manager. A 19-year old long haired metaller, I did precisely what you’d expect: hired a bunch of my mates and rode it like I stole it.

Waking up then was a joy. I could just lie there. The store didn’t open until 11. Nothing was fighting for my attention. I didn’t have the world trying to force its way in through my bedroom window.

Returning to those days in Retro Rewind is a warm bath for my brain. Not just the nostalgia of reliving one of the best jobs I’ve ever had, but for the tranquillity I felt when that was my life.

These days, we’re not so fortunate.

If I’m lucky, I’ll get about five to ten seconds before my memory kicks in. The times we live in feels like waking up with a hangover every morning, only you’re terrified about what everyone else did last night.

When I check my phone, the facesucking begins. Optimised, customised, and more engage than a Star Trek TNG boxset.

It is this dopamine-fuelled hellscape in which I’ve returned to writing my own stuff, arguably two decades later than I should have.

There’s no genuine voices anymore. No one saying anything which reflects on the human condition. Just regurgitated copies of a facsimile, a gutter torrent of beige feeding the engagement beast that is the Pay-per-click Maximiser.

That’s why you’re reading this instead of AI Bayesics #7, which is getting mothballed. In getting in a few writing reps to shake off the dust, I have now been marked as ‘mad AI bastard’ by the algorithm. Last week’s post got 35,000+ impressions. AI Bayesics #6 – a guide to Claude Cowork – got a whole 5 likes.

If you’ve been following it though, don’t fret. Jacob Rackley has recently called me out for being old school and ‘writing things’ so I’m thinking of having a stab as this video nonsense.

Roll with the punches

In my last gig, my vision was simple: uplift the entire organisation with AI. Clean and concise. Just like my video rental store. And – just like the video rental store – it was a façade covering for the labyrinthian nightmare of roles and responsibilities achieving such a thing requires.

Being somewhat of a free radical in both team size and resources has its upsides but also leaves you somewhat prone to Superman Syndrome. This is magnified by understanding the risk of AI. The more you learn about it, the more you feel compelled to work like hell to make sure this isn’t a complete catastrophe.

When I started this journey three months ago, I kept the vision but expanded the scope. I wanted to help everyone.

Over dinner, an investor challenged my utopian ideals with a far more Darwinian take apparently rife in the Valley – more along the lines of ‘if they can’t keep up, fuck ‘em’.

Personally, I’m more of a ‘rising tide raises all boats’ kind of guy. But I also realise now that I’ve been living in a bubble. I am it. I am the AI bubble.

Speaking with a whole range of people since starting Bayes, I’ve encountered a vast spread of AI literacy and understanding. It’s genuinely been one of the difficulties writing on the topic – I have no idea what entry point people are arriving at my pieces at.

AGI’s a great example. I have severely overestimated common understanding of the term and its implications. This poses somewhat of an existential problem for society if the prediction I made back in January that it will arrive by 2028 comes to fruition.

There simply isn’t time. I don’t have the resources. I can’t save everyone. I am not the Son of Krypton.

Valley types are already preparing for their Darwinian banquet. Jeff Bezos recently launched Project Prometheus. Essentially a private equity vehicle with extra vulture beaks, Bezos plans to buy up companies with deep real-world data when they fail to make the jump to AI and use that data to create his own AGI. If Bezos had a moustache, he’d certainly be twirling it.

Personally, I’m borrowing a phrase out of 🌱🤝🌍 Nicolas Sauvage‘s playbook and focusing my effort on supporting impact scalers.

Along this journey, I’ve had venture capital execs I’ve coached tell me I’ve changed their life. I took on a pro bono gig helping a spinal research charity make their first steps into AI - they pretty much bit my hand off and are flying now. I’m supporting an incubator find ways of reducing operational friction with AI.

In all cases, these are folks which work with others to create positive impact in the world. The sort of people I want to see make it across the AI bridge and not end up as a plaything for Bezos and co. It may fall short of my original utopian vision, but it’s enough to chart my course around.

Onwards

On my morning walks, I often go via Plantation Road, the nice part of Leighton Buzzard. I used to do a paper round there and dreamed of one day owning one of the houses. I feel like I was closer to that dream as an 11-year-old paperboy than I am at 41.

My final payout has dwindled from potential first rung on the property ladder to sarcastic laugh from a cash machine. As the pennies get sucked out, the black dog’s barking gets louder. The rock bottom of my mum’s sofa beckons.

Have I made a terrible mistake? Is AI just a tsunami of bullshit that’s washed me away? What if I’m right and we’re all dropping the ball on AI, collectively building our way to catastrophe? What if I starve before any of that matters?

The fear is whipped up by the stress of pulling constant 12 hour days to just keep my head above water. A vortex of anxiety-tipped razor blades constantly howls around my stoic mental defences, looking for any weakness to exploit.

I long for the simplicity of my video rental days. I want to write books, not markdowns or video shorts optimised for content. All I really want in life is some form of stability – a small house, my partner in it, and a dog we can adore. Given the combination of a broken social contract with my generation, an economy in the toilet, and the advent of AI, I doubt it’s something I’ll ever achieve.

And yet, I keep turning up. I’m still putting in the reps. I’m still putting one foot in front of the other.

I started this journey hungover and miserable. It was not by choice. As I leaned into the wind, I knew I’d have to embrace the suck, that it’d be hard and I’d want to quit, that the obstacle was the way.

I’m here for an adventure, not a cruise. Rising to the challenge and pushing my limits hurts like hell, but I keep getting back up and asking for more.

I’ve got no idea how it’ll pan out.

But I know I’ll find out by winging it.

Thank you to the tonne of new subscribers from the last post. Quick intro: I’m Gregg – a large language meatbag and AI strategist. I write about AI risk and living through this weird time we’re all in. If you enjoyed this piece, please do drop a like, leave a comment, and share it with your network.