throwaway9274

10 ms·

31.
by throwaway9274 3y ago
We still don’t know what LLMs are for. By this I don’t mean LLMs aren’t useful. We have known that there was novel behavior happening as early as GPT-2. GPT-3 represented a clearly novel transformative technology. But we still didn’t know
32.
by throwaway9274 3y ago
Maybe, depends on the threshold for human creativity, which is still in flux. You could copyright the underlying character maybe. But its a major headache for these companies and a symptom of the broader fact that copyright needs legislativ
33.
by throwaway9274 3y ago
Everybody’s too concerned about copyright for training data, no one’s concerned enough that the outputs are considered non-copyrightable. $50mm pre-product valuations for startups producing 3D assets for games via prompting a foundation mod
34.
by throwaway9274 3y ago
Early depictions of AI doom are fascinating because the Christian origins of the myth are not yet well-concealed. In the same way that Kurzweil portrays the Heaven of the singularity, I Have No Mouth, and I Must Scream is an excellent Dan
35.
by throwaway9274 3y ago
Here’s what’s going on: -Companies that operate internationally make revenue and incur costs in many different countries. -As a result, they owe taxes to many different national authorities. -Each national authority has rules for how costs
36.
by throwaway9274 3y ago
35%, oops.
37.
by throwaway9274 3y ago
The focus on robotics research starts to make more sense when you dig into the financials. Disney’s FY2022 operating income was $12.7B. $7.9B (!) of that comes from theme parks. That’s 65%. Everything else Disney does (Pixar movies, Disney
38.
by throwaway9274 3y ago
Happy to do the research for you. My consulting rate is $500/hr.
39.
by throwaway9274 3y ago
“It looks gpt-generated.” See, this is an example of a contentless ad hominem. Believability weighting is when you say, “Here’s proof this argument is bad. The community from which this argument originates makes these types of bad arguments
40.
by throwaway9274 3y ago
We are the product of the cultures with which we choose to engage. Some communities have positive signal. Others have negative signal.
41.
by throwaway9274 3y ago
I respectfully disagree. The portion regarding the norms in EA communities argumentation habits is the most weighty part of the analysis. The EMH example is a supporting detail. The thesis is this: EA communities are prone to insularity, hu
42.
by throwaway9274 3y ago
There is a particular flavor of writing common to those whose identity is bound up with the so-called “effective altruist” community and their various online forums. It usually combines a reference to Bayes Theorem, a deep hubris, and a sig
43.
by throwaway9274 3y ago
Definitely one of the benefits of local models. You could theoretically auto-suppress any tokens that lead to a refusal to answer path and re-generate. But you’re probably best off grabbing an existing uncensored chat SFT, like the Llama v2
44.
by throwaway9274 3y ago
There is no way to create a censored model without losing performance. Techniques for PPO and RLHF (the technical basis for the censorship mechanisms) inevitably destroy some parts of the model’s ability to “reason” (as measured by benchmar
45.
by throwaway9274 3y ago
Yes. Meta deliberately let their responsible AI people heavily censor the chat-tuned model. Meta also released the uncensored base model, on which the open source community then performed its own chat fine tunes. This was a canny strategy t
46.
by throwaway9274 3y ago
Sure. But more people is good. More traffic is bad.
47.
by throwaway9274 3y ago
Absolutely right. To situate in terms of recent events, there have been three stages of trying to delay / influence the California CPUC robotaxi expansion approval vote: 1. Tried a politician scare campaign. Failed because it was too t
48.
by throwaway9274 3y ago
Yeah, there’s a weird thing with how they “interpret” system prompts for identity. If you suppress tokens “OpenAI”, “Microsoft”, and “Google”, GPT-4 will tell you that it is Project December, a language modeling project to duplicate the dec
49.
by throwaway9274 3y ago
The Computer Fraud and Abuse Act makes interference with a computer system a felony punishable by five years in Federal prison. That’s what intentionally and maliciously blinding a self-driving car is. That’s not something affected by local
50.
by throwaway9274 3y ago
You’re 100% the same guy under a brand new burner account. This isn’t a movement, and you’re not an activist. Your actions will have knock on effects that harm millions, especially the disabled and elderly. Keep at it, chucklehead. Soon the
51.
by throwaway9274 3y ago
It was literally the District 5 Supervisor Dean Preston’s “youth coordinator” aid who astroturfed this with some DSA people, to distract from his handling of the Tenderloin. He got caught bragging on Twitter and then deleted the tweets. You
52.
by throwaway9274 3y ago
They conducted en masse poisoning of the global population. For what benefit? Non-stick surfaces? How were they supposed to know? How about the deformities in animals around their factory runoff? How about studies prior to mass production
53.
by throwaway9274 3y ago
So the last point is a fair one. Tech companies have a history of skirting laws to achieve market share, then using their customer base to get the law changed. That isn’t what’s happening here though. The driverless car companies don’t have
54.
by throwaway9274 3y ago
“There are no urgent, legit needs that make these things necessary or acceptable.” —Americans alone spend more than 70,000,000,000 hours each year driving. -On average, not counting commercial drivers, each American spends over 13 hours a w
55.
by throwaway9274 3y ago
There is no data to indicate that self-driving cars have caused delays on public transit at any greater rate than human drivers, so that seems irrelevant. Re: relevancy of theft and burglary, addressed in prior responses below. Scapegoating
56.
by throwaway9274 3y ago
Sure, let me grab the links from my bookmarks. Adding below as I go: “What if SF never pulls out of its Doom Loop?” - https://www.ft.com/content/71d8013d-9d94-441e-b2d1-3039c0439... -This one brought the simmering deba
57.
by throwaway9274 3y ago
You’re talking about an industry that could be responsible for a 5-10% GDP boost, freeing up millions for non-driving labor. The revenues to the city where those companies are based would be in the billions.
58.
by throwaway9274 3y ago
Let me be clear that I love it here and will never leave. It’s an amazing city with incredible innovation, a massive tech ecosystem, walkable streets, fantastic parks, easy access to nature. But our leadership has gone totally head-in-the-s
59.
by throwaway9274 3y ago
> Those protests will go the way of the Google Bus protests: make a lot of noise, before they disappear. Totally agree, I expect they’ll dissipate pretty quickly, because there’s no actual committed support. >These have been a mainsta
60.
by throwaway9274 3y ago
This is a bad faith argument. Transit is not down because of self-driving cars, but due to remote work. Cell traffic at city center is dramatically lower in SF compared to pre-pandemic baselines as compared to other major cities. That’s not

More ›