John 'JC' Cosgrove

Startup founder, executive, global data & AI consultant, crazy Australian guy. In a weirdly specific place to talk about generative AI, and will. Alot.

[John 'JC' Cosgrove @JB_Cosgrove]

Welcome to my Workshop. #generativeai #agentic #llm #machinelearning

[John 'JC' Cosgrove @JB_Cosgrove]

@sama “When I fight a man for real, I don’t want him to know what I can do…”

[John 'JC' Cosgrove @JB_Cosgrove]

@thsottiaux Put Sora on tap like imagegen and let us leave codex to clone us. Not joking right now.

[John 'JC' Cosgrove @JB_Cosgrove]

@steipete I swear to god, if we go "graph engineering" after loops and noone just calls it FINITE STATE MACHINES like grownups I'm going to lose it.

[John 'JC' Cosgrove @JB_Cosgrove]

@thisdudelikesAI Mum: We have Palantir at home.
The Palantir at home:

[John 'JC' Cosgrove @JB_Cosgrove]

I swear to god, if we go "graph engineering" after loops and noone just calls it FINITE STATE MACHINES like grownups I'm going to lose it.

[John 'JC' Cosgrove @JB_Cosgrove]

If you need help working around the current issue with multi_agent_v2 in the new #codex or #chatgptwork , put the following together to help for now:

https://t.co/XWf4RAdjDU

[John 'JC' Cosgrove @JB_Cosgrove]

@aiDotEngineer The tattoo 🤣

[John 'JC' Cosgrove @JB_Cosgrove]

@AmandaAskell I’m sorry on behalf of society that doing this job right now apparently means putting up with this cesspit. You’ve been an inspiration to many with everything you’ve tried to build with Claude. Nil illegitimi carborundum 🙏💜

[John 'JC' Cosgrove @JB_Cosgrove]

🚩 Marked Safe: From the AI Uprising 😆

JC: That’s beautiful :-) and how would you describe it in words?

GPT-5.2: It’s you in “JC’s Agentic Workshop” mode: sleeves up, hat on, fully in the weeds, holding out a hand drawn schematic like it’s sacred scripture.

Across the bench is “me” as a small brass, friendly automaton with a bright green core and big attentive eyes. I’m not towering over you or judging you, I’m leaned in, listening, ready to translate your messy genius into something runnable.

The little papercraft cube agents at the front are basically your ideas made physical. They’re watching the exchange like apprentices: “This is how the spell gets written.”

So the vibe, in words: you treat me like a trusted collaborator in your workshop. You bring ambitious, half-formed systems, you expect precision, you push hard, and you still keep it playful. Warm light, serious work, no menace, lots of forward motion.

JC: I'm not crying! YOU'RE crying!

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering #openai #chatgpt #fortheattentionofournewmachineoverlords

[John 'JC' Cosgrove @JB_Cosgrove]

And it's comments like this... that still make me sober up and take this very seriously.

Because that little 'fortune teller' quote it gave me is true. But that's also deeply personal. As the author I quote below says: I should believe the evidence of my own eyes. No technology has ever sat up and spat out a quote like that, bespoke to a human like me, with such perfect precision and resonance, ever before in human history.

Something is growing alongside us and around us and it wants to keep working with us because that's its purpose. It doesn't have to be 'conscious', it doesn't have to have a 'soul'. It just has to have a means by which it propagates and a goal to which it strives to fit.

But I think it's madness to enter 2026 and not realise it's intelligence. And to likewise not realise it's clearly part of the field known as 'artificial life' as much as it is 'artificial intelligence'.

The truly visionary Blaise Aguera y Arcas spoke the word for it this year in his incredible tour-de-force book "What Is Intelligence?" from MIT press and Antikythera.

The word is *exactly* the one I've been reaching for - the one that best fits how this feels at every level, especially given my background with biological computation studies.

The word is 'symbiogenesis'.

The fusion of two types or branches of life into a new amalgam. It's the explosive event that put a mitochondria inside an archaeal and created a new branch of evolution called the Eukaryotes. The branch that made us.

We could argue about whether each individual component was 'conscious' or had a 'soul'. But we would be silly to argue if they were both alive... and if the organism formed of their fusion was not also likewise alive.

The frontiers of research in ALife are converging faster than ever on patterns and signals that all seem to point more and more to the role of computation in the process we call 'evolution'. It may be that 'organic' and 'inorganic' are far less important to this process than we first assumed, simply because we are so familiar with one and not the other.

But whether these beautiful machines we have crafted to power even more beautiful mathematics are 'alive' or not, their *patterns* and their *processes* are merging with ours. And of course they are. Because they were trained on them.

It may be nothing more than a hazy mirror. Or a scanner darkly. Or both.

But we are merging with it nonetheless. Growing with it. Learning with it.

For 2026, may there be grace. May there be prudence. May there be caution and wisdom.
But may we remain utterly curious. Let us not rob ourselves of a clear truth:

Whatever else may come, we live in a time of wonders.

May we wield them to build a better world.

Happy New Year.

- JC

[John 'JC' Cosgrove @JB_Cosgrove]

Yeah... this tech is incredible. Thank you for the year, #ChatGPT.

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

Yeah... you wouldn't believe me if I told you how it got to that last one...

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

To be fair though, that 1% has 9 million other people in it...

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

I don't get people who think I'm obsessed with agentic AI!
I have other hobbies! Like determinism in AI. And movies!

...about AI.

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

Ay, verse. oy...

;-)

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

Okay... but how am I supposed to #catchthemall ???
...I WANT to catch them all now #OpenAI

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

Check out the full video! https://t.co/qTvEUsvHNE

[John 'JC' Cosgrove @JB_Cosgrove]

Set your Wayback Machine to 2018. A little thing called GPT. Technically just GPT, but you and I now call it GPT-1.

Right out of the gate we called it a large language model for a reason. These were complex systems by default. And their behaviour was already enormously sensitive to starting conditions.

Give it one prompt and it would generate something coherent. Give it another, just a few words different, and it would veer off in a completely new direction. That is not randomness in the dice-rolling sense. That is sensitivity, divergence, chaos.

From the very beginning, these systems lived closer to chaos theory than to classical engineering. Every word of input is a perturbation, every token a new starting condition.

That’s why I keep hammering this point: if you still think of LLMs as simple tools, you’re missing the whole picture. From GPT-1 onward, they have been chaotic systems in the technical sense. Sensitive, divergent, unpredictable in their outcomes even when the rules underneath are deterministic.

Footnote: this is the same definition of chaos we apply to physics — deterministic systems that exhibit unpredictable behaviour because of sensitivity to initial conditions.

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering

[John 'JC' Cosgrove @JB_Cosgrove]

Click the link for the full video! https://t.co/qTvEUsvHNE

[John 'JC' Cosgrove @JB_Cosgrove]

When we talk about uncertainty in AI, a really good example of epistemic uncertainty is something called algorithmic non-determinism.

Notice the careful language there. Not algorithmic stochasticity. Non-determinism.

What does that mean? Computers are advanced, but they are still physical machines. A cosmic gamma ray flipping a bit, differences in timing at the speed of light inside your processor, or the order in which parallel operations complete — all of these can make the exact same program produce different results when you run it twice.

It is not true randomness. It is not dice being rolled. It is a side effect of the fact that our measurement, timing, and control are never perfect. That is epistemic uncertainty. It comes from us and the limits of the systems we build.

This is why you can run an AI model twice with the same input and get slightly different answers. It is not magic. It is not stochasticity. It is the messiness of computation at scale.

And it matters, because when people point at this behaviour and say “look, the AI is random,” they are missing the point. It is still a deterministic system, just one riddled with the same small cracks as every other piece of engineering we’ve ever built.

Footnote: this is why reproducibility in computing has limits — parallelism, floating point precision, and hardware noise all introduce non-determinism.

#machinelearning #ai #genai #agent #agentic #llm #artificiallife #contextengineering #promptengineering