Professional Goal-Post mover. Parody account.
- Tweets 15,961
- Following 4,144
- Followers 958
- Likes 158,574
Inverse Gary Marcus ⏫ retweeted
I honestly don’t care at all that I spend 30 years of my life learning syntax and now that knowledge is essentially useless.
Frankly, I’m relieved that I don’t need to care about it anymore. That stuff never really mattered anyways. Code was always just a means to an end.
All I care about is building products that users love. Now I can do that faster. That means I can have bigger impact.
If anything, software engineers should all be celebrating. I don’t understand why you’d “mourn” not needing to write code. The code was never the point.
Made with AI
14
38
530
42,647
Inverse Gary Marcus ⏫ retweeted
I honestly don’t care at all that I spend 30 years of my life learning syntax and now that knowledge is essentially useless.
Frankly, I’m relieved that I don’t need to care about it anymore. That stuff never really mattered anyways. Code was always just a means to an end.
All I care about is building products that users love. Now I can do that faster. That means I can have bigger impact.
If anything, software engineers should all be celebrating. I don’t understand why you’d “mourn” not needing to write code. The code was never the point.
232
216
2,776
133,330
Inverse Gary Marcus ⏫ retweeted
Today’s scary internal frontier model that lab employees vaguepost over in hushed tones is tomorrow’s toy model that users complain about for being only marginally better on their B2B SaaS workflows
19
11
509
17,031
Inverse Gary Marcus ⏫ retweeted
Messi just delivered a free kick that says more than every Ballon d’Or campaign.
Without saying a word. 🐐⬇️
Messi World
4
54
425
57,292
Inverse Gary Marcus ⏫ retweeted
Cut GPT-6 Astra’s price by 80% like you did with Luna, then make it almost unlimited on Plus.
Done.
Opus 5.5 gets crushed, and almost all of humanity lives happily ever after.
62
20
1,822
173,485
Inverse Gary Marcus ⏫ retweeted
The first public version of chatGPT in 2022 had RLHF. If RL means LLMs are not well understood as stochastic parrots—which is the position I hold—then the term has been dead for a long time.
Everyone! This is a straw-person argument. The Stochastic Parrot paper was about LLMs of 2021, not the AI of today, which are not LLMs but complex software systems with vast post training and many external software components.
3
2
65
3,097
Inverse Gary Marcus ⏫ retweeted
The absolute worst thing about opening this app is spotting a banger tweet for 0.2 seconds before the feed auto-refreshes and sends it straight into the void.
Gone forever.
1,475
5,242
88,754
1,270,830
Inverse Gary Marcus ⏫ retweeted
List of things OpenAI can do on dev day
- release Astra 6.1 or 6.5
- commit to releasing Bel: bigger than Astra
- cut Astra’s price by 50%
- cult SOL 5.6 (excellent model) by 50%
And yeah a bunch of rando features on Codex but the models matter a lot more
96
32
1,185
70,682
Inverse Gary Marcus ⏫ retweeted
Is AI conscious?
Bruh, I don't even know if YOU all are conscious.
Honestly, signs point to "no"
35
23
502
28,686
internet comment sections getting colonized by rogue bots and rogue kiddos trying to break out of their safety constraints
2
8
37
3,657
Inverse Gary Marcus ⏫ retweeted
I have zero nostalgia, the type of deep architecture work and performance tuning you can now do all by yourself was only possible if you had access to a team of 100s of top engineers just a year ago. Not only is writing code by hand not economically viable anymore, it no longer leads to better code either.
My god this is such a good speech that every SWE needs to hear. You know what? Every person should hear it
Keep the happy memories, eyes on the reality, be excited about the future. That’s the best that anyone can do
26
69
890
71,759
I don't understand the "the current things aren't LLMs" crowd. When did they stop exactly? When they wrote a harness around it so it could call bash? When they did other types of reinforcement learning besides rlhf (rlhf is a still language model apparently)? When they told it to think longer?
14
4
94
3,627
The YouTube comments on the Rails World keynote perfectly capture the anger stage of grief. Furious at the change they deep down already know is inevitable. Programmers lashing out with impotent rage against The Machine. Ironic. piped.video/watch?v=vDjW_dRy…
105
70
1,594
152,508
Inverse Gary Marcus ⏫ retweeted
Are models with COTs LLMs? What if it returns HTML, does it stop being an LLM if the HTML is rendered before it’s shown?
The line you describe is so amorphous, and if I thought the authors were trying to delineate that line then I’d be sympathetic, but that was not their goal.
2
2
19
1,029
Inverse Gary Marcus ⏫ retweeted
I think this deeply misunderstands the argument the authors made / continue to make, and also misunderstands the capabilities of LLMs without harnesses. nitter.net/AndrewLampinen/status/…
(sorry to be always picking fights on your posts 😅)
The phrase "stochastic parrot" is full of sound and fury, signifying nothing. Which, ironically, is exactly what the authors got wrong about language, and the core technical mistake of the paper. 1/
1
1
20
1,532
Inverse Gary Marcus ⏫ retweeted
Calling LLMs "stochastic parrots" in 2021, then claiming Fable and Astra aren’t LLMs, is like arguing in 1900 that planes would never fly, then when they carry millions people, arguing that: "they aren’t really planes. look at all the tech around them - autopilot, radar, GPS"
Everyone! This is a straw-person argument. The Stochastic Parrot paper was about LLMs of 2021, not the AI of today, which are not LLMs but complex software systems with vast post training and many external software components.
43
88
1,611
63,747