The Machine
That
Learned
To Think
There is a particular kind of hubris that runs through the history of artificial intelligence
— the recurring belief, held by brilliant people in well-funded rooms, that the problem was nearly solved. It never was. And yet, here we are.
What follows is not a comprehensive history. It is a set of moments worth understanding — the turns where everything shifted, quietly or catastrophically, and the field became something new.
The Summer That Named Everything
but the ambition sets the direction for everything that follows.
The First Winter
just waiting.
The Brief Heat of Expert Systems
Rule-based systems flood industry. If this, then that. Companies spend millions encoding human expertise into logic trees. For a moment it seems to work. Then the brittleness shows. The world is not made of rules. The second AI winter follows.
Deep Blue Defeats Kasparov
not because a machine won, but because it won through brute force search, not anything resembling understanding. The question shifts: does it matter how the machine thinks, if it wins?
Hinton Quietly Changes Everything
Geoffrey Hinton and colleagues publish work reviving neural networks under a new name: deep learning. Few outside the field notice. Inside the field, a small group of researchers begin to suspect that this is the thing. They are right.
ImageNet and the Proof
AlexNet enters the ImageNet competition and wins by a margin that should not have been possible. Error rates that had plateaued for years collapse overnight. Deep learning stops being a research curiosity and becomes an industrial imperative. Every major technology company pivots within eighteen months.
Attention Is All You Need
GPT, BERT, the entire large language model era. The title turns out to be literally true.
GPT-3 and the Uncanny Valley of Language
but further out than expected. Something has changed in kind, not just degree.
The Year the Public Noticed
DALL-E, Midjourney, Stable Diffusion. Then ChatGPT in November. A hundred million users in two months. The conversation stops being about capability and starts being about consequence. Regulators, journalists, and philosophers all arrive at the same moment, slightly out of breath.
The Infrastructure War
the chips, the data centres, the energy. AI becomes geopolitical. The research papers keep coming. So do the investment rounds.
The Question Underneath
The systems are capable of more than most people expected, and less than some people fear. What remains unresolved is not a technical question. It is a question about what we are building toward, and who gets to decide. That conversation is only beginning.
The winters are worth remembering. Not as failures, but as corrections — moments when the field was forced to reckon with the gap between what it claimed and what it could actually do. The current moment has no winter in sight. Whether that is because we have finally crossed some threshold, or because the investment is too large to admit another pause, is a question worth sitting with.
At what point in this timeline does the story stop feeling like history and start feeling like the present?
Where were you when you first noticed?
If this resonates, see how to apply it to your own work with the interactive Dispatch agent.
Be first to like this dispatch
