ParodyForMeButNotThee retweeted
Was at the WH Correspondents dinner last night, a rare DC trip for me without a subpoena. On the positive side—was exciting, no one was killed, and ended early. I noted a new litmus for status among the gov’t elite—whether you were whisked away by secret service, or left to fend.
ParodyForMeButNotThee retweeted
I know it sounds bad that the unemployment rate for 16-24-year-olds is more than 10% but it's...normal
The average since 1948 is 11.6%
teapaw.org/NewsWire_US/status/196…
ParodyForMeButNotThee retweeted
ParodyForMeButNotThee retweeted
figma as a product is being squeezed from both edges. its value proposition feels increasingly outmoded.
t2c (text-to-code) is iterating fast with tools like v0 that feel magical, & the gap between conceptual design & implementation is closing. if v0 or another contender can make dev-ready outputs that bypass figma’s static canvas, it’ll be existential. meanwhile, canva’s low-bar simplicity is siphoning off non-power users on marketing use case.
figma’s reliance on craftsmanship design culture *feels* vulnerable in a world moving toward automation & commodification.
ParodyForMeButNotThee retweeted
After 25.3 million autonomous miles driven, @Waymo vehicles have an 88% reduction in property damage claims and a 92% reduction in bodily injury claims compared to human drivers per mile driven. 🚖
Our mission is to be the world's most trusted driver and a new study by @SwissRe validates the exceptional safety record of the Waymo Driver. Learn more: waymo.com/blog/2024/12/new-s…
ParodyForMeButNotThee retweeted
Open-Source LLMs Are Far Safer Than Closed-Source LLMs
While everyone agrees that we don't have AGI yet, a reasonable number of researchers believe that future versions of LLMs may become AGIs
The crux of the argument that closed-source companies use is that it is dangerous to open-source these LLMs because, one day, they may be engineered to develop consciousness and agency and kill us all.
First and foremost, let's look at LLMs today. These models are relatively harmless and generate text based on next-word predictions. There is ZERO chance they will suddenly turn themselves on and go rogue.
We also have abundant proof of this - Several GPT 3.5 class models, including Llama-2, have been open-sourced, and humanity hasn't been destroyed!
Some safety-ists have argued that the generated text can be harmful and give people dangerous advice and have advocated heavy censorship of these models. Again, this makes zero sense, given that the internet and search engines are not censored. Censor uniformly or not at all.
Next, safety-ists argue that LLMs can be used to generate misinformation. Sadly, humans and automated bots are more competent than LLMs at generating misinformation. The only way to deal with misinformation is to detect and de-amplify it.
Social media platforms like Twitter and Meta do this aggressively. Blocking LLMs because they can generate misinformation is the same as blocking laptops because they can be used for all kinds of nefarious things.
The final safety-ist argument is that bad actors will somehow modify these LLMs and create AGI! To start with, you need a ton of compute and a lot of expertise to modify these LLMs.
Random bad actors on the internet can't even run inference on the LLM, let alone modify it. Next, companies like OpenAI and Google, with many experts, have yet to create AGI from LLMs; how can anyone else?
In fact, there is zero evidence that even if future LLMs (e.g., GPT-5) don't hallucinate or mimic human reasoning, they will ever have agency.
The opposite is true - there is tons of evidence to show they are just extremely good at detecting patterns in human language, creating latent space world models, and will continue to be glorified next-word predictors.
Just like a near 100% accurate forecasting model is not a psychic, an excellent LLM is not AGI!
Open-sourcing models have led to more research and a better understanding of LLMs. Sharing research work, source code and other findings promotes transparency, and the community quickly detects and plugs security vulnerabilities, if any.
In fact, Linux became the dominant operating system because the community quickly found and plugged vulnerabilities.
AFAICT, the real problem with LLMs has been research labs that have anthropomorphized them (attribute human-like quality), causing FUD amongst the general population who think of them as "humans" that are coming for their jobs.
Some AI companies have leveraged this FUD to advocate regulation and closed-source models.
As you can tell, the consequences of not open-sourcing AI can be far worse - extreme power struggles, blocked innovation, and a couple of monopolies that will control humanity.
Imagine if these companies actually invented AGI.
They would have absolute power, and we all know - absolute power corrupts absolutely!!
ParodyForMeButNotThee retweeted
There is at least one industry research lab where the leadership believes that (super)human-level AI:
- is attainable
- is a scientific research question, not just a question of more compute and more data.
- is not "just around the corner". It will take a while.
- is not an existential risk.
- requires contributions from the entire research community, because no one has a monopoly on good ideas.
- hence requires open source platforms and open research, because no one has a monopoly on good ideas.
- is going to change human condition for the better.
ParodyForMeButNotThee retweeted
If only OpenAI had remained open-source...
The transparency and open-source attention would have ensured safety.
This mega fight between the safety-ists and capitalists would have been avoided.
Hopefully, this will convince more companies to embrace open-source AI.
ParodyForMeButNotThee retweeted
NOT YOUR WEIGHTS, NOT YOUR MODELS
I hate to make this parallel, because (a) I respect everyone involved and (b) there is zero true wrongdoing alleged, (c) everyone seems to be acting on sincere beliefs, and (d) OpenAI is a historical accomplishment technologically, and regardless of today's events everyone should be proud of what they've done.
But — *solely* from the perspective of the immediate trouble caused to millions of users, investors, and businesses — there's unfortunate similarity to what crypto went through last year with FTX. Because:
1) Billions in value has apparently been destroyed overnight, though we'll need to wait for the next OpenAI valuation to confirm that.
2) Many lawsuits could fly, and the courts will be busy.
3) The EA-driven cultural chokepoint that DC was going to use to centralize the space has seemingly been decentralized overnight.
4) Most importantly, the stability of a centralized service that many users and businesses depended on has been called into question.
On that last topic, we'll have to see what happens to API stability. They could keep it together. Twitter did keep the service up under serious pressure last year, with a lot of people leaving.
But OpenAI feels more abrupt. There could be a lot of IQ and domain knowledge suddenly walking out the door without a transition plan. And Microsoft could turn off compute as leverage, and probably do other things in their contract. Maybe they'd gain access to the weights and ask people to move to Bing AI or something.
We don't know. At a minimum, there is at least huge *uncertainty.*
So, if the lesson of FTX was that we need truly self-sovereign money, the lesson of OpenAI is that we need truly open source AI.
Not your keys, not your coins.
Not your weights, not your models.
ParodyForMeButNotThee retweeted
The employees at OpenAI just lost billions of dollars in secondary share sales that were about to happen at a $90b valuation — that’s over. Done.
I think OpenAI will lose half their employees, the 12-18 month lead, and 90% of their valuation in 2024.
Just insane value destruction
ParodyForMeButNotThee retweeted
The OpenAI board appears to have skipped or sped up due diligence on their new CEO
teapaw.org/eshear/status/16914893…
Replying to @webdevMason
Hmm. Maybe. But between 40-60% of women seem to have rape/non-consent fantasies. Why would you assume it’s not genuine sex fantasy for these girls? Or did I misunderstand and you agree it’s genuine fantasy, you’re just talking about the source of the fantasy?
ParodyForMeButNotThee retweeted
Look, all I know is: the probability of success for most AI-first companies went up by 50%; anyone that knew sama understood his relentlessness
i don't know what happened
ParodyForMeButNotThee retweeted
The instability at OpenAI is yet another reason for open source AI - do you really want to have so much power concentrated in the hands of one or two organizations, where whoever is at the helm (or whoever plots behind their backs to seize power) wields so much control and power via AI systems?
ParodyForMeButNotThee retweeted
Today if you were to withdraw enough money to cover the first month, last month and security deposit for the average NYC 1 bedroom from your bank account, your bank will be forced to file a Suspicious Activity Report - previously called a 'criminal referral form' - with FinCen.
ParodyForMeButNotThee retweeted
There isn't a shred of evidence that AI poses a paradigmatic shift in safety.
It's all fantasies fueled by popular science fiction culture suggesting some sort of imaginary but terrible, terrible catastrophic risk.
ParodyForMeButNotThee retweeted
There’s no point in using GPT-3.5 in production — the API has frequent downtime, plus high response latencies lead to subpar end user experiences.
Instead, I’ve found a Mistral 7B finetune deployed with VLLM to be 10x faster(!), way cheaper, and *stupidly* simply to self-host.
ParodyForMeButNotThee retweeted
I wrote in June 2021 how Sam Altman’s departure from YC had been under explored
ParodyForMeButNotThee retweeted
GOP operative found guilty of funneling Russian money to Donald Trump washingtonpost.com/dc-md-va/…

































