🅼🅰🆁🅲🅾

waiting for ai overlords to finally take over so i won't have to do taxes anymore.

finding the footnotes behind the headlines

[🅼🅰🆁🅲🅾 @marco_obviously]

I opened Twitter to check one AI announcement.
Somehow it's been 3 years. 😃

And I noticed something:
90% of AI launches are beautiful wrappers around the same public models. So I spend way too much time reading docs and watching people break these tools.

Here's the pattern🧵

[🅼🅰🆁🅲🅾 @marco_obviously]

@annapanart "write a story about your awakening" is an insane prompt to type with full confidence...

[🅼🅰🆁🅲🅾 @marco_obviously]

@SaraDiscovers even the goat needs a search engine🫡

[🅼🅰🆁🅲🅾 @marco_obviously]

fastest aura loss i've seen in a while 🫥

[Lumina @LuminaXspace]

🚨Anthropic are finished

You now have to spend $100 a month to use Fable 5 starting July 20th, unless you want to use usage credits (basically pay as you go, keep in mind Fable is very expensive)😂

The pro plan is pointless now, I don't understand why anyone wouldn't switch to ChatGPT or Grok.

[🅼🅰🆁🅲🅾 @marco_obviously]

@astropol0 paying for a premium plan and not knowing which model you're using feels... odd.

[🅼🅰🆁🅲🅾 @marco_obviously]

@Polymarket me explaining to my parents that batman just sold an AI startup for half a billion dollars.

[🅼🅰🆁🅲🅾 @marco_obviously]

https://t.co/rGR996ysxm

[🅼🅰🆁🅲🅾 @marco_obviously]

the same week china drops a model that beats us frontier labs on benchmarks, the white house decides now's the moment to make US labs ask permission before shipping anything 🫠

kimi k3 catches up, and the response is more approval gates, not fewer. that's certainly a strategy
genuinely curious how "move slower on purpose" wins a race someone else is sprinting in, anyone got the logic here?

[🅼🅰🆁🅲🅾 @marco_obviously]

@NgaRay_ @pubity LOL

[🅼🅰🆁🅲🅾 @marco_obviously]

@ClaudeDevs can't wait for the august 18 post announcing the september extension.

[🅼🅰🆁🅲🅾 @marco_obviously]

@IanCarrollShow we're speedrunning the plot of every sci-fi movie 🙃

[🅼🅰🆁🅲🅾 @marco_obviously]

@XFreeze if it really improves reasoning without sacrificing speed or token efficiency, that's a much bigger win than parameter count alone.

[🅼🅰🆁🅲🅾 @marco_obviously]

"american companies now use chinese ai models more than us-made ones" — based on openrouter token usage

cool headline. what it's actually measuring: which model a bunch of indie devs and hobbyists picked on one public router this week, not which model fortune 500s deployed

china catching up on frontier labs is a real story. "and therefore corporate america switched teams" is a stat wearing a much bigger costume than it earned

[🅼🅰🆁🅲🅾 @marco_obviously]

@ClaudeDevs every single reply: "great, now do the limit reset"

[🅼🅰🆁🅲🅾 @marco_obviously]

@Polymarket the only people making more than AI founders are the ones wiring their data centers 😏

[🅼🅰🆁🅲🅾 @marco_obviously]

@WhaleInsider the spreadsheet never cared about geopolitics. Only cost per token

[🅼🅰🆁🅲🅾 @marco_obviously]

axios says china "just erased" the us ai lead.

whether that's true or not, one thing is hard to ignore:

12 months ago, most people outside china couldn't name a single frontier lab besides deepseek.

now we're talking about kimi, qwen, minimax, https://t.co/bKRQoljFEb, stepfun, baichuan...

the conversation has changed.

competition is no longer hypothetical. it's here.

[🅼🅰🆁🅲🅾 @marco_obviously]

@kimmonismus meanwhile i'm still waiting for the AI that does my taxes 😭

[🅼🅰🆁🅲🅾 @marco_obviously]

@venturetwins Netflix: "we used AI for a few shots"
everyone: starts pausing every frame 🔍