Welcome, humans.

Four years ago, ChatGPT launched and Corey and Grant basically had the same reaction: oh no. The writing was rough. The implications were rougher.

So for this episode, we did something we have weirdly never done before: walked backward through the last four years of AI, about six months at a time, and asked what counted as “mind-blowing” at each stop.

Reasoning. Voice. Long context. Open models. Coding agents. Computer use.

The funniest part is how many “old” “miracles” now feel like table stakes.

Here’s our favorite parts:

• (48:31) The ChatGPT launch-day existential crisis: Corey remembers asking it to write an article, getting something mediocre back, and immediately deciding he needed to become “the hardest to fire guy in writing and publishing.”

• (13:33) When agents finally felt like agents: Earlier automation still made you prebuild the workflow. The late-2025 shift was being able to tell a system what you wanted and watch it move between tools and actually do the job.

• (29:11) Safety vs. open access: Grant argues that treating those as opposites is a false choice. The goal should be systems that are safe by design and still broadly available.

• (44:50) The first time AI wrote code in front of Corey: Code Interpreter opened Python, built a way to inspect a file, and permanently changed his mental model of what these systems could do.

• (1:14:21) “Machine labor on demand”: Grant’s long-term prediction is that people may eventually allocate compute and robotic labor like capital, then team up to build things instead of thinking of themselves primarily as labor.

The whole exercise makes one thing painfully obvious: humans are bad at feeling exponential change while we are living inside it. Every six months, the baseline moves, then your brain quietly decides the new thing was always normal.

Why watch this? Because this is less a model-history lecture and more a reset for your intuition. If you want to understand where AI might go next, it helps to remember how absurdly much moved from “impossible” to “Tuesday” in four years.

Watch and/or Listen now: YouTube | Spotify | Apple Podcasts

P.S. At 38:00, Corey makes the case that voice mode was one of the most underappreciated AI shifts of the whole timeline. Once AI could live in the car, kitchen, or your glasses, the interface stopped being “a website you visit.”

Keep scrolling for the five shifts that changed what AI actually feels like to use, our November 18 San Francisco event, and the last two podcast episodes.

Special shout out to Zoominfo for sponsoring this episode.

Additional Resources: The five shifts that changed the baseline

No need to memorize every model name in the episode. The useful part is seeing how the interface kept widening.

• Better answers: Early ChatGPT and GPT-4 were mostly about making the text output more coherent, useful, and trustworthy.

• More context + tools: By 2023, models could hold much more information and start using things like Python, files, and other tools inside the reasoning loop.

• Multimodal AI: Voice, vision, and images turned the chatbot into something that could see, hear, and respond through more than text.

• Reasoning + open competition: Reasoning models and systems like DeepSeek R1 pushed the frontier in a different direction: spend more compute thinking, and make more of that capability available outside a few closed labs.

• Agents: The final jump was from “answer my question” to “go do the work.” Coding agents, browser / computer use, and long-running orchestration started turning the model into an operating layer.

And underneath all of it is the technical idea that made the modern era possible: the 2017 Transformer paper, “Attention Is All You Need.” Grant’s closing challenge is the one worth keeping: look at the last four years, then imagine the next four.

Which begs the question: with all this amazing new machine intelligence, and general purpose machine labor from robots on the horizon…

…what kind of world do we want to create?

UPCOMING EVENT: The Neuron IRL in San Francisco

The Neuron is going IRL in San Francisco on November 18 starting at 4:30 PM PT. If you’re in town, come hang out with us for drinks, bites, and a live recording of The Neuron podcast. We’ll have The Neuron community together in one room, plus some special guests and plenty of AI conversation.

Huge thanks to Alumni Ventures, Slack, and Sonar for helping us bring the night to life. It’s free, but space is limited, so RSVP and save your spot.

🎙️ In Case You Missed It…

Two recent interviews and episodes we think you’ll love.

1. Napster Is an AI Company Now. Seriously.

TL;DW: Napster CEO John Acunto explains why the company pivoted from music into visual AI agents and what he calls “streaming intelligence.” His thesis is that businesses should use agents to learn directly from customers while keeping control of the resulting data and models.

Why you should watch: Come for the “wait, Napster does WHAT now?” hook. Stay for a surprisingly useful conversation about AI interfaces, memory, customer-data ownership, edge compute, and why John thinks all of this eventually points toward robots.

Watch / Listen: YouTube | Spotify | Apple Podcasts

2. Robots That Learn Jobs by Watching You

TL;DW: Standard Bots CEO Evan Beard explains how industrial robots are moving beyond rigid point-by-point programming. Teams can increasingly demonstrate a job, let the system learn from examples, and use vision to handle variation that would break a traditional robot workflow.

Why you should watch: This is the practical version of “AI + robotics.” Less humanoid theater, more factory machines that normal people can teach and that keep working when the real world gets messy.

Watch / Listen: YouTube | Spotify | Apple Podcasts

Have a topic you want to learn about? Request it here!

Subscribe to our YouTube Channel for more!

Subscribe on YouTube to help us bring in more builders, researchers, and guests who can teach you something useful about AI every week.

Stay curious,

The Neuron Team

P.P.S. Love the newsletter, but don’t want these podcast and livestream announcement emails? Don’t unsubscribe. Adjust your preferences to opt out of them here instead.