Joined April 2022
- Tweets 688
- Following 354
- Followers 3,068
- Likes 1,397
116 Photos and videos
Pinned Tweet
4 Oct 2023
First new Substack piece in quite some time — "The question concerning technology in 2023"
Some reflections on our book several months later, future directions, the ultimate destiny of our society and what we can do to shape it. 🐝
harmlessai.substack.com/p/th…
1
2
10
6,137
8 Sep 2024
Why would anyone sign up to work for one of these e/acc founders memeing about going “founder mode”?
These are men whose ego and masculinity is entirely tied up in them running a business. It’s certain that power tripping, irrational business decisions will result. Red flag.
1
1
12
1,256
6 Sep 2024
Thanks to everyone who came to THE ANTI SINGULARITY PICNIC last weekend
12
776
H̶armless retweeted
5 Sep 2024
Why is every AI critique like “AI sucks at everything but ALSO it’s too powerful and it will take over everything and we have to prevent that from happening”? If it sucks then it probably won’t take over. If it takes over then it probably doesn’t suck.
5
2
26
2,369
20 Sep 2023
Some people are very performative about telling you they're "not impressed" by GPT-4 because it is only about as smart and as accurate as the average eighth grader.
And surely, given how there has been a lull in dramatic progress the past few months, it will stay that way.
2
10
2,895
H̶armless retweeted
20 Sep 2023
A good sign for AI Harmony
A bad sign for AI Alignment
20 Sep 2023
Probably nothing, but Poolside AI is the 2nd notable startup to be announcing moving its HQ from the US to Paris in a week? (other one being The Browser Company)
🇺🇸🛫🇫🇷
1
1
2
2,576
20 Sep 2023
Right now there is no school of thought that accepts that there is a danger of AI hostile to human values — but rather than asking the government to prevent this, fears that the government is the body most likely to create it in the first place.
3
2
4
1,379
19 Sep 2023
Why do people pretend like they can predict when human-level AI will arrive, quibbling over their "timelines"?
When has anyone in the past ever been able to predict when any technological development will happen? Has this ever worked?
6
1
15
1,769
15 Sep 2023
To identify as an "accelerationist" is an admission of powerlessness before the sweeping, conquering path of planetary technology.
The fact that CEOs and VCs are now identifying like this is a sign that they no longer understand what they are bringing into this world.
3
4
38
2,611
14 Sep 2023
We live in what is sure to be the greatest century in man's history, and are coming to age right now in the midst of a universal crisis in all forms.
The pace of technological development all but ensures that some of us living now will participate in a posthuman breakthrough.
1
18
1,091
14 Sep 2023
And yet, a collapse of all social forms ensures that this future will be seized only by those able to navigate a radically indeterminate flux of meaning and shape it for themselves.
The age of Zarathustra
Acts of creation will be possible the likes not seen since ancient times.
1
2
14
798
14 Sep 2023
This is an odd, pathetic, stupid decision from Valve.
One of the best use cases for AI in its current state is that it can automate making some of the art assets in a video game so a smaller team can make something ambitious. But arbitrary rage of Luddites sadly disallows this.
6 Sep 2023
Another steam game rejected for "AI" except they didnt use AI. If valve even has the suspicion of you using AI, they will reject.
Can you even really tell if somethings made by AI 100% of the time?
3
2
44
3,218
12 Sep 2023
Professors of machine learning are sadly the type to tweet this. He thinks language is about what is "useful" to communicate, as if we are all signals division officers in a military.
I told my wife I love her 7 times this morning.
These people know nothing about intelligence.
10 Sep 2023
Twitter bios like Obama's ("dad, husband, President, citizen"') are an annoying piece of virtue signaling. Saying you're a dad and husband conveys almost no information. Say something useful instead.
3
23
2,238
1 Sep 2023
Harmless AI was lucky to make an appearance on Barrett Avner's Contain, and we continued to fire our various shots at Yudkowsky, the e/acc crowd, and nearly everyone else out there who discusses AI with insufficient self-consciousness.
Check it out! 🐝
New episode 168. with @realityspammer
on the bizarre world of X-Risk, @harmlessai, Corporate Surrealism, E/ACC b*llsh*t, more, up in bio
1
3
15
3,328
23 Aug 2023
A lot of people on X app have forgotten how to construct original thoughts. People regurgitate clichés, desperate for the approval of peers. The language has been flattened to the genericness of an RLHFed LLM, without the need for AI.
Twitter is rapidly becoming Reddit.
3
4
32
2,164
23 Aug 2023
By contrast, D&G — quite interestingly — blame totalitarian regimes as emerging from the opposite phenomenon: the establishment of massive informational-redundancy in speech. Under National Socialism, one is expected to repeat "Heil Hitler" over and over greeting all in a room.
1
1
9
691
23 Aug 2023
Novel creative subcultural movements cannot emerge from the "velocity" of a mass herd repeating a cliché together, no matter how good it may feel to join a swarm in the moment. To seize a better future, we must figure out how to actually express ourselves.
1
12
693