Ethan

I teach you how to actually use AI. Real tools, real workflows, no hype — one practical tip at a time.

[Ethan @gambletan]

https://t.co/oR6huP6EKL

[Ethan @gambletan]

Most people skip this step when using AI coding agents. Ask it to explain its approach out loud BEFORE it writes code. You'll catch massive logic errors before they become code. What's your biggest frustration with AI coding tools so far?

[Ethan @gambletan]

You're using ChatGPT wrong if you only ask it to create things. Ask it to critique what you've written instead. Much better at editing than generating. What's something you've written that needs a second set of eyes?

[Ethan @gambletan]

Create a hostile subagent in Claude Code. Paste: "Create a subagent whose only job is to be hostile: critique everything I write, find weaknesses, challenge assumptions." It wants to destroy — not please you like your main agent. What's your best code-attacking prompt?

[Ethan @gambletan]

Most people don't know AI can read screenshots and pull out data.
Screenshot a table, a receipt, a chart — paste it in and ask questions.
What have you used this for?

[Ethan @gambletan]

https://t.co/HX4WAG7xWy

[Ethan @gambletan]

Fable 5 went dark overnight.
Your app probably broke.
You built on someone else's floor.
The floor moved.
Do you own your inference?

[Ethan @gambletan]

Stop renting intelligence from providers who hate your margins.

#AI

https://t.co/RD8QM3b92n

[Ethan @gambletan]

Real-time speech translation just got AlignAtt4LLM.
Cascaded ASR + MT with "forced alignment."
The demos work perfectly in the lab.
But every production deploy hits the latency wall.
You can't align what the user already stopped listening to.

#AI #LLM

[Ethan @gambletan]

Cyera wants 80x revenue at $12B.
Meanwhile every CISO knows their data is already breached.
They're not buying security.
They're buying compliance theater.

#AI #Microsoft

[Ethan @gambletan]

Scorsese supports AI in film.
Every studio exec just got permission.
The auteurs are aligning with the machine.
Soon every script gets the algorithm treatment.
Who controls the creative stack now?

#AI #LLM

[Ethan @gambletan]

The Grammys are banning AI music.
Meanwhile every label uses AI to A&R.
They're gatekeeping the output while automating the input.
The humans picking winners are already gone.

#AI #LLM

[Ethan @gambletan]

Spent a weekend on ollama's "stuck at 99%" bug — open 2 years, 124 comments, no fix. The maintainer correctly diagnosed it as a Cloudflare/R2 stall. That diagnosis is why nobody fixed it. The actual bug is in the client watchdog, 5 lines: https://t.co/QjenkeziGx 9c51 Tag:
@simonw @awnihannun @ollama

[Ethan @gambletan]

Algorithmic hiring creates perfect circles of bias.

Three vendors control 80% of resume screening.
Same training data trains every filter.
Same unconscious biases get encoded everywhere.

[Ethan @gambletan]

Anthropic just hit profitability.
Companies are shocked at their LLM bills.
You built on someone else's meter.
The meter just started running.

#Anthropic #LLM

[Ethan @gambletan]

The New York Times unionizes against AI monitoring.
Meanwhile every media company tracks writers with AI.
They're not fighting surveillance.
They're negotiating the price of being watched.
Local newsrooms don't have unions.

[Ethan @gambletan]

Law schools debate AI accountability.

Enterprises ship AI agents to production with zero liability frameworks.

The code fails silently.
The human takes the blame.
The API provider has immunity clauses.

Who's liable when your AI agent empties the wrong bank account?

#AI

[Ethan @gambletan]

AI agents fail silently in production.

Enterprises track server crashes.
They don't track logic drift.

Your agent worked yesterday.
Today it hallucinates customer data.

You'll find out when the lawsuit hits.

#AIAgents #LLM

[Ethan @gambletan]

San Francisco nonprofits replace volunteers with robots.

The humans stopped showing up.

The robots didn't.

Who controls the kitchen when charity goes automated?

#AI #AIAgents

[Ethan @gambletan]

Google's new "anything-to-anything" AI model lets you deepfake your kid's stuffed animal.

The demo worked perfectly.

Now every parent will try it.

Half will get banned for violating terms they never read.

Would you trust your family photos to their content policy?

#Google