Just Killing Time
TL;DR
This was Dylan Curious’s first solo livestream, and the whole thing was basically an accidental Q&A — he went live because a time-zone mixup left him with an hour to kill before a stream with Wes Roth and guest Julia McCoy, then ended up talking with about 35 viewers from places like NYC, Ethiopia, Missouri, and the Philippines.
He framed his AI use less as productivity-maxing and more as curiosity with better vibes than TikTok — his examples were hilariously mundane but revealing, like asking AI who the world’s most famous celebrity hairstylists are or using it to chase etymologies, obscure definitions, and random tangents.
He’s genuinely bullish on vibe coding and thinks coding agents may be the real strategic center of the AI race — he cited companies shifting from humans doing 80% of the work to Claude doing 80%, mentioned a joke app that made $5,000 by making a laptop moan when slapped, and riffed on recursive AI building data centers as the real win condition.
He used Sora’s shutdown to sketch a bigger market thesis about OpenAI’s priorities — his read is that meme-video products are fun but strategically less important than coding and enterprise, and he even floated Disney as a plausible buyer that could fold Sora into Disney+ and unlock it with full IP rights.
His most serious point was that privacy is probably gone, so governance matters more than secrecy — he used the example of wireless tire-pressure sensors being trackable via tiny identifiers and argued the real civic challenge is deciding how governments and companies are allowed to use unavoidable data.
The stream also showed what makes his channel distinct: he prefers obscure, thoughtful sources over speed-chasing headlines — he said he skips AI news everyone else is already covering, leans on Medium, blogs, email lists, and Patreon tips, and would rather stay curious than burn out trying to win the news race.
Summary
A First Livestream Born From a Time-Zone Mistake
Dylan opens in full “am I even live?” mode, checking the mic, reading the first chat message, and admitting this is the first truly live thing he’s ever done on his own channel. The reason is almost too perfect: he had an hour to kill before going live with Wes Roth and Julia McCoy, thanks to a time-zone issue, so he figured he’d just see if anyone showed up.
AI as a Better Way to Waste Time
Very quickly, the stream settles into Dylan’s natural register: curiosity as a lifestyle. He says he mostly uses AI to ask endless side-quest questions — definitions, etymologies, weird facts, and random searches like which celebrity hairstylists cut famous people’s hair — and laughs that it’s probably still healthier than TikTok.
Vibe Coding, Clawbot Envy, and the New Coding Meta
The chat pulls him into coding tools, and he sounds both excited and slightly left behind, joking that everyone has a “Clawbot” except him. He talks about his own game startup moving toward Godot, reacts to Claude’s new computer-use features, and brings up a vibe-coded novelty app — slap the laptop, it moans — that reportedly made $5,000, which clearly blows his mind.
Sora Is Gone, but Maybe Disney Should Buy It
When the topic shifts to Sora, Dylan’s reaction is half nostalgia, half relief. He says AI is too powerful to waste entirely on “meme machines,” but then spins out a surprisingly concrete idea: Disney could buy Sora for a few billion, keep or rename the brand, wall it off with Disney IP, and turn it into a Disney+ native creative engine while OpenAI keeps pushing multimodal video elsewhere.
Why His Channel Feels Different
One of the most revealing stretches is when he explains how he covers AI news. He says he intentionally skips stories everyone else is already racing to post, instead pulling from email lists, blogs, Medium, and long-form opinionated sources because he’d rather find the weird, interesting angle than optimize for speed — and that’s also why he doesn’t feel burned out the way other creators do.
AI D&D, Bees, and Other Side Quests That Somehow Work
The middle of the stream gets delightfully chaotic in a very Dylan way. He riffs on AI for Dungeons & Dragons, imagines Google Genie making whole interactive fantasy worlds in real time, gets excited about using AI as a game master, and then detours into a sincere mini-monologue about bees as a kind of collective brain — “when a bee dies you lose like a neuron” — before pausing to ask Claude what an apiary is.
Privacy Is Probably Over, So Citizenship Matters More
The conversation turns serious when he talks about privacy and AI. His core argument is that in a world of pattern-matching systems, true privacy is basically impossible — he uses the example of tiny wireless tire-pressure sensors in cars being trackable by roadside AI receivers — so the real question is not whether data exists, but who gets to use it and under what constraints.
Adderall, Identity, and the Bigger AI Race
Near the end, Dylan gets unusually personal, talking about spending roughly 10 years on prescribed Adderall, how it made coding and data work feel rewarding, and how strange it is to realize part of his old identity was chemically reinforced. That leads into a bigger AI thesis: coding models matter because recursion matters, and whoever gets AI to improve code, robots, and data-center construction in a self-reinforcing loop may effectively “win” the race.
Wrapping Up With a Global Audience Surprise
As he prepares to jump to the Wes Roth stream, he realizes people are watching from all over — not just the US, but Ethiopia and the Philippines too — which genuinely seems to change how he thinks about going live. He signs off still half-surprised that 35 people showed up, promising he might do more of this as long as it doesn’t crowd out the structured videos he actually loves making.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
The Retirement Email Isn't a Warning
Model retirements now arrive every few weeks; the config-eval-rehearsal loop turns each deprecation email from a fire drill into an afternoon swap.
Playbook
The Cheapest Model That Passes
OpenRouter lists 400 models behind one API. The fix for choosing isn't a better leaderboard, it's a four-step protocol that ends in a real eval.
Playbook
Cheap Models, Hard Tasks
Most agent workflows route every step to the frontier model by default. The bill scales with how chatty the agent gets, even when most steps don't need that brain.