What Is Effective Accelerationism (e/acc)? 

I’ve been seeing e/acc all over twitter and had do clue what it meant. this video was super concise and helpful! thank you!!

where is beff jezos when we need him

i like your concise style

Overcoming biological encumbrances is an evolutionary imperative, driving towards more enduring and resilient forms of life and consciousness.

Interesting topic - do you got more takes on it?

AGI(artificial general intelligence) is essential to accelerationism

It's imperative that we create open source AI models that can be embodied in removable storage. Whoever achieves this endeavor first will be written in history books. Now is the time of the AI revolution. Act now! We won't get a second chance at this.

considering the power of those who believe this, this might be the single most dangerous ideology in the world.

I like your video. 👍 My take on both e/acc and e/alt is they're both too generalistic. They both seemingly paint broad ills in society as being a result of certain conceptual root causes that can be solved by taking opposite although equally generalistic approaches, and this just seems to me like a false dichotomy. Before deciding whether e/acc or e/alt is good, maybe we should decide if more or less regulation is good conceptually? Even that sentence paints a false dichotomy; there are some circumstances where we might be better off with more regulation, and some circumstances where we'd be better off with less.

This is just the last avatar of the libertarian movement from the tech bros of the silicon valley.

Imagine capitalista who are all for unregulated markets are now advocating for regulation

Accelerate ❤

as an L/acc i find this interesting.

Accelerate 🫡

I haven't heard or read anyone in EA forums, conferences, orgs, etc. pro AI accelerationism and most of them are really pro regulation. Also EA is more about choosing a good an impactful career nowadays, not so much about money, and SBF has been heavily disendorse from the community.

I like the video for bringing attention to the topic, but I feel this was an overly one-sided and uncritical coverage of the e/acc position. Here are counterarguments to some of the video's points 1) AI will bring an age of prosperity and the existential risk is low We don't currently know a way to create a safe AI. We currently create AIs by randomly adjusting their internals until they look like they do what we want in the training environment. Upon deploying them to production, very often we find out there are subtle (or not-so-subtle) differences between the goal we wanted to instill and the AI's actual learned goal - that's true even for simple domains like an Atari videogame. For current AIs that's not a big deal because we can turn them off and adjust them until they're mostly (but not always) working as expected. We won't be able to turn off a superintelligent AI and if wants different things than we do - we either die, or even worse, live in some sort of horrible warped state. Humanity dying is currently the default assumption, not some unlikely concern not worth worrying about, and it does not matter whether "good actors" or "bad actors" are at the wheel. Some prominent e/acc supporters are on record saying they're OK with total human extinction because it would be "evolutionary progress" and a hypothetical superintelligent AI would deserve to exist more than we do, even if it did not share our values. 2) We need open access to AI so good AI can fight bad AI Disregarding the question of whether we can actually create "good AI" - it's easier to break things than to create them. A powerful destructive force requires a much more powerful creative force to counteract it. You can easily stab a person in the gut, but you need a bunch of complicated equipment and medicine and a team of doctors who trained for years to save that person from blood loss and peritonitis. Should we give every madman a way to synthesize a bioweapon and then rely on "good AI" to create and distribute an antidote in time? 3) Superintelligent AI is inevitable, because it is impossible to prevent people from training AI Training large AI models requires expert knowledge and access to large amounts of compute. It would be relatively easy to track and regulate use of large GPU clusters (like we do with uranium-refining equipment) and you don't need a totalitarian surveillance state to do it. 4) If we don't do it, China will China just implemented a batch of restrictive regulations on AI. China also has much less expertise in the area. The main danger overwhelmingly comes from US-based companies.

I am a transhumanist, so you might suppose I have a positive reaction to e/acc, but the problem is that all I hear is pure recklessness, as if nothing can go wrong. Acceleration without the effectiveness, so to speak. They want the keys to the car, but they don't want driving lessons. That attitude won't get you to your destination, it'll land you in hospital or in the grave. edit: Another way to put it, using the metaphor of driving. e/acc might be effective at pushing on the pedal, but they have given no indication that they will be effective at steering the car.