from the welcome-future-dystopian-AI-overlords dept.
Things are moving at lightning speed in AI Land. On Friday, a software developer named Georgi Gerganov created a tool called "llama.cpp" that can run Meta's new GPT-3-class AI large language model, LLaMA, locally on a Mac laptop. Soon thereafter, people worked out how to run LLaMA on Windows as well. Then someone showed it running on a Pixel 6 phone, and next came a Raspberry Pi (albeit running very slowly).
If this keeps up, we may be looking at a pocket-sized ChatGPT competitor before we know it.
[...]
For example, here's a list of notable LLaMA-related events based on a timeline Willison laid out in a Hacker News comment:
- February 24, 2023: Meta AI announces LLaMA.
- March 2, 2023: Someone leaks the LLaMA models via BitTorrent.
- March 10, 2023: Georgi Gerganov creates llama.cpp, which can run on an M1 Mac.
- March 11, 2023: Artem Andreenko runs LLaMA 7B (slowly) on a Raspberry Pi 4, 4GB RAM, 10 sec/token.
- March 12, 2023: LLaMA 7B running on NPX, a node.js execution tool.
- March 13, 2023: Someone gets llama.cpp running on a Pixel 6 phone, also very slowly.
- March 13, 2023, 2023: Stanford releases Alpaca 7B, an instruction-tuned version of LLaMA 7B that "behaves similarly to OpenAI's "text-davinci-003" but runs on much less powerful hardware.
Related:
DuckDuckGo's New Wikipedia Summary Bot: "We Fully Expect It to Make Mistakes"
Robots Let ChatGPT Touch the Real World Thanks to Microsoft (Article has a bunch of other SoylentNews related links as well.)
Netflix Stirs Fears by Using AI-Assisted Background Art in Short Anime Film
Paper: Stable Diffusion "Memorizes" Some Images, Sparking Privacy Concerns
The EU's AI Act Could Have a Chilling Effect on Open Source Efforts, Experts Warn
Pixel Art Comes to Life: Fan Upgrades Classic MS-DOS Games With AI
Related Stories
Last night, a Reddit user by the name of frigis9 posted a series of six images that feature detailed graphical upgrades to classic MS-DOS computer games such as Commander Keen 6 and The Secret of Monkey Island. The most interesting part is how they did it: by using an image synthesis technique called "img2img" (image to image), which takes an input image, applies a written text prompt, and generates a similar output image as a result. It's a feature of the Stable Diffusion image synthesis model released last week.
[...] Art quality in image synthesis currently requires much trial and error with prompts and cherry-picking to achieve the kinds of results frigis9 posted—likely hours of work. But with some incremental advances in image synthesis techniques and GPU power, we could imagine an emulator upgrading vintage game graphics in real time within a few years.
The EU's AI Act could have a chilling effect on open source efforts, experts warn:
The nonpartisan think tank Brookings this week published a piece decrying the bloc's regulation of open source AI, arguing it would create legal liability for general-purpose AI systems while simultaneously undermining their development. Under the EU's draft AI Act, open source developers would have to adhere to guidelines for risk management, data governance, technical documentation and transparency, as well as standards of accuracy and cybersecurity.
If a company were to deploy an open source AI system that led to some disastrous outcome, the author asserts, it's not inconceivable the company could attempt to deflect responsibility by suing the open source developers on which they built their product.
"This could further concentrate power over the future of AI in large technology companies and prevent research that is critical to the public's understanding of AI," Alex Engler, the analyst at Brookings who published the piece, wrote. "In the end, the [E.U.'s] attempt to regulate open-source could create a convoluted set of requirements that endangers open-source AI contributors, likely without improving use of general-purpose AI."
In 2021, the European Commission — the EU's politically independent executive arm — released the text of the AI Act, which aims to promote "trustworthy AI" deployment in the EU as they solicit input from industry ahead of a vote this fall, EU. institutions are seeking to make amendments to the regulations that attempt to balance innovation with accountability. But according to some experts, the AI Act as written would impose onerous requirements on open efforts to develop AI systems.
In a recent example, Stable Diffusion, an open source AI system that generates images from text prompts, was released with a license prohibiting certain types of content. But it quickly found an audience within communities that use such AI tools to create pornographic deepfakes of celebrities.
But out of 300,000 high-probability images tested, researchers found a 0.03% memorization rate:
On Monday, a group of AI researchers from Google, DeepMind, UC Berkeley, Princeton, and ETH Zurich released a paper outlining an adversarial attack that can extract a small percentage of training images from latent diffusion AI image synthesis models like Stable Diffusion. It challenges views that image synthesis models do not memorize their training data and that training data might remain private if not disclosed.
Recently, AI image synthesis models have been the subject of intense ethical debate and even legal action. Proponents and opponents of generative AI tools regularly argue over the privacy and copyright implications of these new technologies. Adding fuel to either side of the argument could dramatically affect potential legal regulation of the technology, and as a result, this latest paper, authored by Nicholas Carlini et al., has perked up ears in AI circles.
Related:
Getty Images Targets AI Firm For 'Copying' Photos
Over the past year, generative AI has kicked off a wave of existential dread over potential machine-fueled job loss not seen since the advent of the industrial revolution. On Tuesday, Netflix reinvigorated that fear when it debuted a short film called Dog and Boy that utilizes AI image synthesis to help generate its background artwork.
Directed by Ryotaro Makihara, the three-minute animated short follows the story of a boy and his robotic dog through cheerful times, although the story soon takes a dramatic turn toward the post-apocalyptic. Along the way, it includes lush backgrounds apparently created as a collaboration between man and machine, credited to "AI (+Human)" in the end credit sequence.
[...] Netflix and the production company WIT Studio tapped Japanese AI firm Rinna for assistance with generating the images. They did not announce exactly what type of technology Rinna used to generate the artwork, but the process looks similar to a Stable Diffusion-powered "img2img" process than can take an image and transform it based on a written prompt.
Related:
ChatGPT Can't be Credited as an Author, Says World's Largest Academic Publisher
90% of Online Content Could be 'Generated by AI by 2025,' Expert Says
Getty Images Targets AI Firm For 'Copying' Photos
Controversy Erupts Over Non-consensual AI Mental Health Experiment
Microsoft's New AI Can Simulate Anyone's Voice With Three Seconds of Audio
AI Everything, Everywhere
Microsoft, GitHub, and OpenAI Sued for $9B in Damages Over Piracy
Adobe Stock Begins Selling AI-Generated Artwork
AI Systems Can't Patent Inventions, US Federal Circuit Court Confirms
Last week, Microsoft researchers announced an experimental framework to control robots and drones using the language abilities of ChatGPT, a popular AI language model created by OpenAI. Using natural language commands, ChatGPT can write special code that controls robot movements. A human then views the results and adjusts as necessary until the task gets completed successfully.
The research arrived in a paper titled "ChatGPT for Robotics: Design Principles and Model Abilities," authored by Sai Vemprala, Rogerio Bonatti, Arthur Bucker, and Ashish Kapoor of the Microsoft Autonomous Systems and Robotics Group.
In a demonstration video, Microsoft shows robots—apparently controlled by code written by ChatGPT while following human instructions—using a robot arm to arrange blocks into a Microsoft logo, flying a drone to inspect the contents of a shelf, or finding objects using a robot with vision capabilities.
To get ChatGPT to interface with robotics, the researchers taught ChatGPT a custom robotics API. When given instructions like "pick up the ball," ChatGPT can generate robotics control code just as it would write a poem or complete an essay. After a human inspects and edits the code for accuracy and safety, the human operator can execute the task and evaluate its performance.
In this way, ChatGPT accelerates robotic control programming, but it's not an autonomous system. "We emphasize that the use of ChatGPT for robotics is not a fully automated process," reads the paper, "but rather acts as a tool to augment human capacity."
Not to be left out of the rush to integrate generative AI into search, on Wednesday DuckDuckGo announced DuckAssist, an AI-powered factual summary service powered by technology from Anthropic and OpenAI. It is available for free today as a wide beta test for users of DuckDuckGo's browser extensions and browsing apps. Being powered by an AI model, the company admits that DuckAssist might make stuff up but hopes it will happen rarely.
Here's how it works: If a DuckDuckGo user searches a question that can be answered by Wikipedia, DuckAssist may appear and use AI natural language technology to generate a brief summary of what it finds in Wikipedia, with source links listed below. The summary appears above DuckDuckGo's regular search results in a special box.
[...] Update (March 9, 2023): We spoke with a representative of DuckDuckGo and they said they're using OpenAI's GPT-3.5 and Anthropic's Claude as LLMs. "We're experimenting with OpenAI's recently announced Turbo model, too," they said.
Related:
Robots Let ChatGPT Touch the Real World Thanks to Microsoft (Article has a bunch of other SoylentNews related links as well.)
(Score: 3, Funny) by Opportunist on Thursday March 16 2023, @07:15PM (10 children)
I didn't say you couldn't, I said you shouldn't.
(Score: 2, Funny) by Anonymous Coward on Thursday March 16 2023, @08:48PM (9 children)
Interesting. Why do you say I shouldn't?
(Score: 0) by Anonymous Coward on Thursday March 16 2023, @10:00PM (1 child)
That's what the luddites say
(Score: 0) by Anonymous Coward on Friday March 17 2023, @12:26AM
Stop asking ChatGPT what to say.
(Score: 4, Touché) by maxwell demon on Friday March 17 2023, @04:51AM (6 children)
Eliza, is that you?
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Tork on Friday March 17 2023, @05:31PM (5 children)
🏳️🌈 Proud Ally 🏳️🌈
(Score: 2) by coolgopher on Saturday March 18 2023, @01:04AM (4 children)
I wonder where the "tso" in the name from?
Sb = Sound Blaster
ai = artificial intelligence
tso = ???
(Score: 2) by Tork on Saturday March 18 2023, @01:09AM (3 children)
🏳️🌈 Proud Ally 🏳️🌈
(Score: 2) by coolgopher on Saturday March 18 2023, @01:11AM (2 children)
Hmm! Do you know for a fact, or "merely" an educated deduction? /curious
(Score: 2) by Tork on Saturday March 18 2023, @01:35AM (1 child)
🏳️🌈 Proud Ally 🏳️🌈
(Score: 2) by coolgopher on Saturday March 18 2023, @05:17AM
Thanks, I appreciate it!
(Score: 4, Insightful) by Snotnose on Thursday March 16 2023, @08:52PM (6 children)
Doesn't mean it would be a pleasant experience, but I could.
A couple months ago I spent a couple quality hours with ChatGPT while imbibing. Oops, I meant imbibing whilst dorking with ChatGPT cuz I was too drunk to watch The Kardashians. My conclusion? ChatGPT is a very accomplished bullshit generator. Pretty much everything I asked it gave results straight from Wikipedia, with a dash of right/left wing bullshit tossed in to make me think it was reinforcing my beliefs when, in fact, I was drunkenly playing the part of Lou Dobbs or Al Sharpton when I asked my last question.
The secret to success is to never run from hard work. A brisk walk usually suffices.
(Score: 0) by Anonymous Coward on Thursday March 16 2023, @09:07PM
> The funnel will be held tomato.
This. Also Auto-Complete (from another SN link, forgot to where), texting the baby sitter:
Can you sit on my face.
(Score: 4, Funny) by Gaaark on Friday March 17 2023, @11:54AM
Man: I'd HAVE to be drunk to watch The Kardashians!
--- Please remind me if I haven't been civil to you: I'm channeling MDC. ---Gaaark 2.0 ---
(Score: 3, Insightful) by Freeman on Friday March 17 2023, @02:02PM (3 children)
I posit that running Doom on a more recent Raspberry Pi could be a very pleasant experience. At least as far as running and playing Doom like it was meant to be played. The Raspberry Pi is a very capable little computer and just because it doesn't run fast, doesn't mean it's not good. A raspberry pi doesn't need to boot in 3 seconds, finish loading the OS/bloatware in 10 seconds more, and load Doom in 5 seconds. We're very spoiled with modern instant gratification. The Raspberry Pi loads pretty quickly and something like DOSBox is likely what you want to run ancient games like Doom on modern hardware anyway.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Insightful) by Freeman on Friday March 17 2023, @02:04PM (2 children)
Also, a lot of older windows programs just don't work on modern versions of Windows. In fact the likes of Civilization II is extremely hard to actually get to "mostly function" on a modern Windows system. Whereas, if you load the game in WINE on a Linux computer, you can make it work like it was meant to be played. Complete with loading the music and wonder videos from a CD.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 0) by Anonymous Coward on Saturday March 18 2023, @02:37AM (1 child)
Really? Is it extremely hard for you realize you can run Linux in a VM on Windows? 🤣
(Score: 2) by Freeman on Monday March 20 2023, @06:10PM
WINE is the necessary ingredient. In the event that you can get WINE to work on Windows, natively or in a VM on a Linux guest, sure.
The extremely hard comment was looking for a trust-worthy hacked EXE or other hacky way of getting it to run on a modern computer (probably illegal, but the copyright holder seem to be ignoring that) to run Civ2 natively on Windows 10 (for example). First of all, the original uses 16-bit architecture and second of all, even after getting it to "mostly function" there were bits that just didn't work right. Whereas, with Linux and WINE, it was a breeze. Yes, setting up Linux in a VM isn't terribly difficult and could be a way to do it. Considering I setup the thing for my Dad who knows nothing about computers. I wouldn't want to support Windws + VM + Linux + game. Linux + Game == Good, especially since I supplied the computer anyway.
Sad thing is that after I got CivII running perfect and CivIV running perfect, he tried CivII and decided that nostalgia had been the driving force behind wanting 2 vs 4. So he still pretty much exclusively plays Civ IV. I did get him one of the newer Civ games, I think Civ V, but he was much happier with Civ IV. Personally, I think CivIV is the sweet-spot of the genre. Lots of modern conveniences, with fairly quick turns. Civ V and Civ VI turns just take way too long. When I hit pass, the computer should be able to take their turn(s) very quickly. Not a 10-30 second pause, every time you hit end turn at the very beginning of a game. Which would only get worse once they actually have a bunch of units, cities, etc. Late game with lots of units, cities, etc. always take more time, but the Computer's Turns shouldn't take forever.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 5, Funny) by legont on Thursday March 16 2023, @09:18PM (3 children)
Cool, but where do you get a Pi?
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 3, Funny) by coolgopher on Friday March 17 2023, @08:13AM
Try asking LLaMA?
(Score: 2) by Freeman on Friday March 17 2023, @02:09PM (1 child)
You buy an "Ultimate" package from one of the vendors or pay scalper prices on e-bay. At least with the "Ultimate" package you get a few things other than just the bare board for $100+.
https://www.canakit.com/raspberry-pi-4-extreme-aluminum-case-kit.html [canakit.com] ($144.95 - $184.95) 1GB version in-stock, 2GB & 4GB versions available for pre-order (4gb version shipping in march), and 8GB version out-of-stock.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 3, Interesting) by legont on Thursday March 23 2023, @03:13AM
Since I already own about a dozen of older PIs, I have all those things galore.
I do need a newer PI to play 265 encoded videos, GPT be damned.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 4, Interesting) by DadaDoofy on Thursday March 16 2023, @09:47PM
Does this mean you can run the software against your own data or just process requests locally against a central data repository?
(Score: 0) by Anonymous Coward on Thursday March 16 2023, @11:36PM (1 child)
Not so much "run it" as "crawl it".
Eventually.
(Score: 0) by Anonymous Coward on Friday March 17 2023, @12:28AM
As long as they can bill you for the "tokens," it doesn't matter.
(Score: 1) by Runaway1956 on Friday March 17 2023, @12:36AM (4 children)
Let me download and install an AI like ChatGomePhilTom. Do I get the source code? Well, personally, it doesn't really matter, because I can't read source well enough to decide if there is a back door or not. But, the point is, do I get the source code? I compile it myself? Or, is it an executable blob, and I'm just trusting the source? But, it's an AI, right? Who programmed it's ethics, exactly? Maybe calling home is the ethical thing to do?
I'll pass. I don't want my chatbot informing the FBI and NSA of the locations of my nuclear arms caches. Or how much money I'm extorting from Hunter Biden, or . . . most of you should get the idea here.
A MAN Just Won a Gold Medal for Punching a Woman in the Face
(Score: 3, Informative) by Freeman on Friday March 17 2023, @02:22PM (3 children)
https://github.com/tatsu-lab/stanford_alpaca [github.com] (Sounds like it's open-source with all the stuff you need to roll-your-own. With a possibility of them releasing the "secret sauce", if Meta says they can. Meta's "secret sauce" was already leaked and widely distributed.)
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2, Informative) by guest reader on Friday March 17 2023, @06:42PM (2 children)
Alpaca is "just" a fine-tuning of a LLaMA model.
There is also another open source fine-tuning trainer ChatLLaMA [github.com].
Both are based on pre-trained LLaMA models which means that you will still need to fill in Meta's form to obtain the LLaMA’s weights. Pre-trained LLaMA models have restrictive license: do not share, do not sue, nonpermanent, non-commercial use etc.
The training of LLaMA is otherwise described in research paper LLaMA: Open and Efficient Foundation Language Models [facebook.com]. The training of 65B model took 21 days on 2048 A100 GPU cards. This article was a part of the LLaMA announcement [facebook.com].
(Score: 3, Informative) by coolgopher on Saturday March 18 2023, @01:08AM (1 child)
Actually, you do not need to fill out their form; it's available via bittorrent as well [github.com].
(Score: 3, Interesting) by guest reader on Saturday March 18 2023, @07:21AM
Danger, Will Robinson.
The BitTorrent link is just a pull request from a random guy. This pull request is not merged to the official LLaMA Meta Research repository.
The official page [github.com] still contains the following instructions: