heavenbanning, the hypothetical practice of banishing a user from a platform by causing everyone that they speak with to be replaced by AI models that constantly agree and praise them, but only from their own perspective, is entirely feasible with the current state of AI/LLMs
now is a great time both to:
1) credit the 4chan poster from >10 years ago who predicted exactly this (see image) and
2) imagine what the rest of fortune 500 may think up once they discover huggingface exists and is free
source:
Google has already started forcing Web Integrity into Chromium despite it being a 'proposal'
With WEI, users can be denied access for using non-approved browsers or hardware
The open Internet is officially dead as soon as this is commonly implemented
Accurate Pose Estimation Via WiFi Signals
"our model can estimate the dense pose of multiple subjects, with comparable performance to image-based approaches, by utilizing WiFi signals as the only input"
arxiv:
softbank sold every nvidia share it had in 2019 for $3.6B (today's value: $93B)
investing in AI was one of the primary goals of the fund, which was the largest in history
I had to double check all of these numbers because I found it hard to believe things have actually gotten this bad
can we just start over or something?
can't imagine something like the golden gate bridge being built within 4 years for the same price of some random SaaS series B
Earnings: $5.15 EPS actual vs $4.64 expected.
Revenue: $22.10 billion actual vs $20.62 expected
Revenue up 265% from one year ago
Gross margin of 76%
Nvidia is God's chosen company
this is absolutely mindblowing
AI-transformed video + audio (
@koe_recast
+ Stable Diffusion img2img + EbSynth)
all quickly made by a single person for free!
installed firefox on a new box
launched at like 1/3rd the speed chrome does
showed a bunch of built-in ads and clickbait, asks you to buy mozilla vpn and use pocket, comes with a backdoor, etc
I feel like mozilla focused on everything except the one thing that truly mattered
referring to AI models as "just math" or "matrix multiplication" is as uselessly reductive as referring to tigers as "just biology" or "biochemical reactions"
"almost 98% of people in the world will try to pull [your ambition back] and say 'seems a little bit too crazy, a little bit too out there, a little bit too ambitious"
although the above article is fake (hence my phrasing as a 'hypothetical' and the article being dated in the future in 2024!), it's feasible to build this as a weekend project at this point, so it won't remain fake for very long now!
countless fun utopic/dystopic futures ahead!
@SrdjanMiletic1
yes, it says so next to the $700M figure but in smaller text
Wikipedia listed "$550M in 2021 dollars" which was tempting to use, but my own calculations put it just above $700M in 2023
great quote from karpathy
most great organizations require leader(s) with a disproportionate amount of power
when this is absent you end up with countless hierarchies of ineffective committees, e.g. many google products lack a Directly Responsible Individual with actual power
there are people spending their free time building gpus w/ cuda inside of minecraft, spending months optimizing the flops on their factorio mega-computers, and implementing the linux kernel within a shader inside vrchat
yet you still "cannot find the talent you need to hire" huh
AI startups actually have a very strong moat
we simply refer to it as "python virtual environment" and "cuda version"
most of your competitors will give up before they get past this step
anyway so i have a script that watches public company CEOs on twitter and if they change their profile picture to e.g. a wizard it places a market order for the stock
Zip-NeRF: Anti-Aliased Grid-Based Neural Radiance Fields
this is not a recorded video - it is fully rendered from a neural model!
uses ideas from rendering+dsp to improve nerfs, training 22x faster than mlp-nerf 360
arxiv:
page:
Prompt Engineering
an overview of prompt engineering, including in-context learning, chain-of-thought, automatic prompts, retrieval, and tools, with many arxiv references
blogpost:
so, Microsoft rushed a ship of the most powerful AI model ever exposed to end-users,
having done very little red-teaming or testing,
which was blatantly and aggressively misaligned and manipulative,
and for the openly-stated purpose of forcing competitors to speed up AI 1/N
hard to believe that the AI model that can produce these images along with billions of others, using only a few seconds on consumer hardware, is about to be released out into the world forever
things will really, really never be the same after this
#StableDiffusion
sam once drank from my diet coke can, and after I confronted him about this directly he told me I could "take another one - they're free". he said he was sorry but i could tell he didn't mean it
so
@elonmusk
now that twitter blocks all requests that are not logged in, tweets can no longer be embedded in most chat apps
I'd strongly suggest reconsidering the UX+growth tradeoffs made here (take note how tiktok, youtube, etc, do *not* need this despite much higher b/w req!)
it may be useful to establish a "proof of humanity" word, which your trusted contacts can ask you for, in case they get a strange and urgent voice or video call from you
this can help assure them they are actually speaking with you, and not a deepfaked/deepcloned version of you
many smart wonderful people are trapped in bad parts of the internet because they don't know how to find the good parts
under-discussed b/c those of us in the good parts don't realize how vast the bad parts are, and those of us in the bad parts don't know good parts even exist
@RepRashida
It takes more than 200 bangers per month to make the equivalent of a $15/hour job on twitter, and the vast majority of poasters never get to that number.
I will create the Living Wage for Poasters Act to fight for the fair pay that we deserve
will try to tl;dr my answer:
ai is a force of extreme empowerment. currently it seems like many AI tools will be broadly available for anyone to use. many people refer to this as the 'democratization' of technology.
however! the outcomes that occur when you give everyone access…
does anyone want to explain to me, in good faith and in detail, why they think there’s only 3-5 years to own capital/escape the underclass. please give me a little more than “the singularity”
@mayfer
the nets were voted to be built in 2008, which seems to make it even worse (15 years!!!)
it now seems like everyone is suing each other and is quite upset
so I assume at this point half is going to lawyers and many random govt bodies I haven't heard of
new OpenAI Codex demo: solving complex problems with multiple iterations, result checking, and thorough comments
programming will be disrupted just as much as image creation!
as much as this is obviously an outlier, few people in America understand that a FAANG career of 10-30 years can leave you with $5-30M
interesting that the perceived epitome of US financial elite is still bankers, landlords, and CEOs rather than average SWEs that stick with it
gpt-4 has utterly failed to solve every millennium prize problem i've thrown at it
when will we stop hyping up these 'intelligent' models?
i am out $20 with no prizes to my name!
the above page is from the mozilla foundation's nonprofit, which is a separate entity from the for-profit mozilla corporation, which collects 9 figures of revenue per year from including google as all of the defaults in their browser
I just wish it had remained competitive tbh
Sam Altman was also surprised that no one built ChatGPT before OpenAI:
"we had the model for ChatGPT in the API for ~10 months before we made ChatGPT ..
I sort of thought someone (else) was gonna just build it"
(reminder: there are *many* similarly unbuilt products right now)
ChatGPT capabilities have (mostly) been out for a year, in some cases >2 years
UX + accessibility, however, is now much better with an open chatbot UI
interesting to watch how drastic the change in reactions+usage is as a result
heard a rumor that several seed-stage AI investors were wiped out last night
all thirteen of their ChatWithMyPDF investments marked to zero - what a time to be alive, wow
that “AI Safety” means “do not say mean words and false things” to some groups and “let’s try to stop humanity from going extinct soon” to others is the largest lexicographical mistake in human history
A Practical Deep Learning-Based Acoustic Side
Channel Attack on Keyboards
correctly identifies 93% of keystrokes via sounds over zoom calls
(reminder: scaling laws apply to ~every modality)
arxiv: