#ML

Today’s AI News & Comment

Age of AI: Everything you need to know about artificial intelligence
https://techcrunch.com/2023/06/09/age-of-ai-everything-you-need-to-know-about-artificial-intelligence/

Aberdeen AI trial helps doctors spot breast cancers
https://www.bbc.co.uk/news/technology-65821552

Struggling Meta showcases new AI tools at company meeting
https://www.theguardian.com/technology/2023/jun/08/meta-facebook-ai-mark-zuckerberg

Google lays out its vision of securing AI
https://www.techradar.com/news/google-lays-out-its-vision-of-securing-ai

Meta’s first generative AI feature will be AI stickers in Messenger
https://www.theverge.com/2023/6/8/23754485/meta-generative-ai-stickers-in-messenger

COPYWRITER FIRED AFTER BOSSES STARTED CALLING HER "CHATGPT"
https://futurism.com/the-byte/copywriter-fired-chatgpt

Apple CEO Tim Cook says he uses ChatGPT and is excited about the chatbot
https://www.businessinsider.com/apple-ceo-tim-cook-uses-chatgpt-excited-about-chatbot-2023-6

The 5 best AI chatbots of 2023 (so far)
https://www.businessinsider.com/apple-ceo-tim-cook-uses-chatgpt-excited-about-chatbot-2023-6

THE BACKSTORY TO THE EATING DISORDER BOT THAT WENT OFF THE RAILS IS TOO STUPID FOR WORDS
https://futurism.com/the-byte/backstory-eating-disorder-bot-stupid

The Role of AI in Web3 Marketing
https://hackernoon.com/the-role-of-ai-in-web3-marketing

An AI Robot Learns Recipes by Watching Human Chefs
https://www.psychologytoday.com/us/blog/the-future-brain/202306/an-ai-robot-learns-recipes-by-watching-human-chefs

Microsoft doesn't want us to be scared of AI - but is it doing enough?
https://www.techradar.com/news/microsoft-doesnt-want-us-to-be-scared-of-ai-but-is-it-doing-enough

Bing’s chatbot now lets you ask questions with your voice on desktop
https://www.theverge.com/2023/6/9/23755754/bing-voice-chat-desktop-microsoft

Teaser’s AI dating app turns you into a chatbot
https://techcrunch.com/2023/06/09/teasers-ai-dating-app-turns-you-into-a-chatbot/

Google’s Bard AI can now write and execute code to answer a question
https://arstechnica.com/google/2023/06/googles-bard-ai-can-now-write-and-execute-code-to-answer-a-question/

Nvidia’s AI software tricked into leaking data
https://arstechnica.com/gadgets/2023/06/nvidias-ai-software-tricked-into-leaking-data/

What’s being built in generative AI today?
https://techcrunch.com/2023/06/09/whats-being-built-in-generative-ai-today/

How AI art killed an indie book cover contest
https://www.theverge.com/2023/6/9/23752354/ai-spfbo-cover-art-contest-midjourney-clarkesworld

U.S. Congress to consider two new bills on artificial intelligence
https://www.reuters.com/technology/us-congress-consider-two-new-bills-artificial-intelligence-2023-06-08/

#AI #ArtificialIntelligence #ML #MachineLearning
#LLM #ChatGPT #BingChat #SwarmsOfRampantAIChatbots

Get started with the open-source Amazon SageMaker Distribution 👉 Quickly experiment on your local environment and easily promote them to jobs on SageMaker https://aws.amazon.com/blogs/machine-learning/get-started-with-the-open-source-amazon-sagemaker-distribution/ #AWS #MachineLearning #ML #OpenSource

Researchers find loopholes in #Nvidia's NeMo Framework, which helps developers work with #LLMs, letting them bypass safety restraints to reveal private data.

https://t.co/xdwBnWVIYi

#oops #AI #ML #GenerativeAI #MachineLearning

@danie10 This is cool but imagine what's possible...

If you can train a #ML / #AI to learn how to hack (sort of ugly) images to be readable as QR codes, how about training #ML and generative AI to create beautiful images that nearly invisibly embed data readable only by an #ML trained app?

Outline: Pick from fixed set of 1600 "QRmojiis". Generator ML perturbs image to embed data in pleasing way. Reader ML recognizes base image, takes the diff & reads the perturbations as data.

I really think people who missed Devs from a few years ago really need to go watch it. Much of what it explores feels spot on for the belief and drive for some folks in #AI and #ML at a belief level. Don’t worry it is not killer AI robots. Hulu should resurface this show promo wise stat!

This a broad synopsis of the premise with only minor spoilers if you are curious from the toot but need a little more
https://www.refinery29.com/en-us/2020/03/9505258/what-is-devs-fx-hulu-about-plot#

Mark Zuckerberg: AI will be built into all of Meta’s products https://www.artificialintelligence-news.com/2023/06/09/mark-zuckerberg-ai-built-into-all-meta-products/ #meta #ai #ml #news #tech #technology

Extending a photo with #ML/#AI, this is a fascinating head-to-head between DALL-E & Photoshop generative fill via #Petapixel https://petapixel.com/2023/06/08/which-can-extend-a-photo-better-dall-es-outpainting-or-photoshops-generative-fill/

Introducing popularity tuning for Similar-Items in Amazon Personalize 👉 To give you more control over the degree to which popularity influences Similar-Items recommendations https://aws.amazon.com/blogs/machine-learning/introducing-popularity-tuning-for-similar-items-in-amazon-personalize/ #AWS #MachineLearning #ML

The Khan Lab School trials Khanmigo, an #AI #chatbot developed by Khan Academy to simulate one-on-one tutoring, one of the first such experiments in the US.

https://www.nytimes.com/2023/06/08/business/khan-ai-gpt-tutoring-bot.html

#GenerativeAI #ML #MachineLearning

As someone who has excitedly read about, thought about, and sometimes written about #robots for most of his life, I truly didn't expect interactions with #ML-based bots to give me the #willies.

System1 is hiring a Senior Machine Learning Engineer in #Remote! Apply today at https://www.datasciencejobsusa.com/jobs/senior-machine-learning-engineer-system1--bb156. #datasciencejobs #datascientistjobs #datascience #aijobs #mljobs #machinelearning #engineeringjobs #Hiring #HiringAlert #USA #ML #AI

@bruno_j_navarro

BINGO!

Watched the video of Satya's #AI pitch to the #WSJ from #Davos last year. It was pathetically embarrassing.

All current iterations of #AI weren’t;
• introduced
• launched
• “widely adopted by consumers”
– they were injected.

#AI #ML #BigTech
#Microsoft
#SIliconValley
#BigTech
#AIBlitzkrieg
#OligarchBlitzkrieg
#PeterThiel
#ElonMusk
#SamAltman
#OpenAI
#ChatGPT
#EthicalAI
#AIEthics
#Democracy

I've been distracted with the post-apocalyptic smoke situation, that I missed that AlphaDev discovered faster sorting and hashing algorithms #ai #ml #reinforcementlearning
https://www.deepmind.com/blog/alphadev-discovers-faster-sorting-algorithms

Just added this page to my blog: a "No AI/ML use clause".

https://www.didiermary.fr/a-propos-blog-personnel/no-ai-ml-use-clause/

Probably toothless, but at least, it exists...

#AI #LLM #ML

Starting to think Zuck misses longs walks with Steve so much he's using Tim Cook language (‘products’) to compensate

FaceBook / Meta is a services company and a hela-creepy one at that.

#Meta #Facebook
#AI #ML #XR #VR #MR
#Privacy #Surveillance
#SurveillanceCapitalism
#SiliconValley #BigTech
#ArtificialIntelligence
#GenerativeAI
#AIBlitzkrieg
#OligarchBlitzkrieg
#Democracy

Today’s AI News & Comment

Search engines don't always help chatbots generate accurate answers
https://www.theregister.com/2023/06/07/search_engines_make_chatbots_inaccurate/

UK to host global AI 'safety measure' summit in autumn
https://www.bbc.co.uk/news/technology-65834085

Scientists claim >99 percent identification rate of ChatGPT content
https://www.theregister.com/2023/06/08/scientists_ai_recognition/

New Adobe Express Includes Firefly AI and Better Integration with CC Apps
https://petapixel.com/2023/06/08/new-adobe-express-includes-firefly-ai-and-better-integration-with-cc-apps/

Google Cloud Enterprise Search is getting even better with AI models
https://www.techradar.com/news/google-cloud-enterprise-search-is-getting-even-better-with-ai-models

Adobe opens up its Firefly generative AI model to businesses
https://www.theverge.com/2023/6/8/23753564/adobe-firefly-enterprise-generative-ai-express-commercial

ChatGPT on iOS gets improved iPad support and Shortcuts integration
https://www.theverge.com/2023/6/8/23753570/chatgpt-ipad-app-apple-shortcuts-support-openai

Segway partners with Drover AI, Luna to bring computer vision to e-scooters
https://techcrunch.com/2023/06/07/segway-partners-with-drover-ai-luna-to-bring-computer-vision-to-e-scooters/

Meta plans to put AI everywhere on its platforms
https://www.axios.com/2023/06/08/meta-ai-zuckerberg-announcement-generative

As the AI industry booms, what toll will it take on the environment?
https://www.theguardian.com/technology/2023/jun/08/artificial-intelligence-industry-boom-environment-toll

DALL-E vs Photoshop: Which is Better for Extending Photos?
https://petapixel.com/2023/06/08/which-can-extend-a-photo-better-dall-es-outpainting-or-photoshops-generative-fill/

Ron DeSantis Campaign Makes Trump Attack Ad With AI Imagery
https://petapixel.com/2023/06/08/ron-desantis-campaign-makes-trump-attack-ad-with-ai-imagery/

AI startup Cohere, now valued at over $2.1B, raises $270M
https://techcrunch.com/2023/06/08/ai-startup-cohere-now-valued-at-over-2-1b-raises-270m/

LinkedIn is the next social network to offer AI-powered tools for ad copies
https://techcrunch.com/2023/06/08/linkedin-is-the-next-social-network-to-offer-ai-powered-tools-for-ad-copies/

#AI #ArtificialIntelligence #ML #MachineLearning
#LLM #ChatGPT #BingChat #SwarmsOfRampantAIChatbots

Maggie Appleton (@maggie) gave a fascinating Causal Islands🏝️ talk on how the web's Dark Forest may continue to expand with the introduction of generative AI. How can we express ourselves, form genuine human connections, and share our knowledge when AI agents are able to churn out massive amounts of content while appearing indistinguishable from humans? Maggie explores these thoughts and more: https://youtu.be/VXkDaDDJjoA #generativeAI #AI #ML #LLM

Was able to do an interview on The Futurisity Porcast with Vlad! Here is a link to the show https://podcasts.apple.com/us/podcast/the-futurisity-podcast/id1606606283?i=1000616152841 #podcast #ml #datascince #DataPrivacy

🔖 The Top 5 Papers About #mlops You Should Know (Part 3/4)

Not just another list of #ml, #datascience, #deeplearning papers.

3️⃣ Machine Learning: The High-Interest Credit Card of Technical Debt by D. Sculley, Gary Holt, Daniel Golovin, Eugene Davydov,
et al. (https://research.google/pubs/pub43146/)

4️⃣ The ML Test Score: A Rubric for ML Production Readiness and Technical Debt Reduction By Eric Breck, Shanqing Cai, Eric Nielsen, Michael Salib, D. Sculley (https://research.google/pubs/pub46555/

UK will host global AI summit to address potential risks https://www.artificialintelligence-news.com/2023/06/08/uk-host-global-ai-summit-address-potential-risks/ #ai #ml #uk #europe #ethics #society #government #politics #news #tech #technology

[review] On the Design Fundamentals of Diffusion Models: A Survey
https://arxiv.org/abs/2306.04542

Diffusion models are generative models, which gradually add and remove noise to learn the underlying distribution of training data for data generation. ... we organize this review according to their three key components, namely the forward process, the reverse process, and the sampling procedure.

#ML #MachineLearning #DiffusionModels #persagen #SyntheticMedia

@dahukanna You can take a shortcut with those #probability #calculations and just round it to 0. It's definitely very very close to 0.

I think decades of seeing superhumanly #smart #computers in #SciFi and pop culture are becoming really dangerous now. People naively assume that this is going to be possible now.

It's garbage in, garbage out! If I only have old, definitely very biased #human #decisions to learn from, where is the #unbiased wisdom supposed to come from?!

It's gonna be "but the #computer said so" all the time.

Who said "machine learning is like money laundering for #bias"? Ah,Maciej Cegłowski aka pinboard

🤬

#machinelearning #artificialintelligence #ml #ai #aiethics

Also, #Google #DeepMind blog: #AlphaDev discovers faster sorting algorithms https://www.deepmind.com/blog/alphadev-discovers-faster-sorting-algorithms #ML

Hmmm ... 🤔

"[...] algorithms such as sorting or hashing are used trillions of times on any given day [...] #AlphaDev discovered small sorting algorithms from scratch that outperformed previously known human benchmarks."

Faster sorting algorithms discovered using deep reinforcement learning https://www.nature.com/articles/s41586-023-06004-9 #ML

Join us for the next week's #GIScience research seminar talk, which links #ML with geostatistics✨️
All info ⬇️

System1 is hiring a Senior Machine Learning Engineer in #Remote! Apply today at https://www.datasciencejobsusa.com/jobs/senior-machine-learning-engineer-system1--7a7c0. #datasciencejobs #datascientistjobs #datascience #aijobs #mljobs #machinelearning #engineeringjobs #Hiring #HiringAlert #USA #ML #AI

Today’s AI News & Comment

OpenAI CEO has no IPO plan due to 'strange' company structure
https://www.reuters.com/technology/openai-ceo-has-no-ipo-plan-due-strange-company-structure-2023-06-06/

How to develop artificial super-intelligence without destroying humanity
https://www.theguardian.com/news/audio/2023/jun/07/how-to-develop-artificial-super-intelligence-without-destroying-humanity

‘What should the limits be?’ The father of ChatGPT on whether AI will save humanity – or destroy it
https://www.theguardian.com/technology/2023/jun/07/what-should-the-limits-be-the-father-of-chatgpt-on-whether-ai-will-save-humanity-or-destroy-it

AI needs a regulatory ecoystem like your car, not a Czar
https://www.theregister.com/2023/06/07/ai_needs_a_regulatory_ecoystem/

New Research Finds People Prefer News Videos Made by Humans Over AI
https://petapixel.com/2023/06/07/new-research-finds-people-prefer-news-videos-made-by-humans-over-ai/

Google launches new learning and consulting offers help enterprises on their AI journey
https://techcrunch.com/2023/06/07/google-launches-new-learning-and-consulting-offers-help-enterprises-on-their-ai-journey/

Google’s generative AI support in Vertex AI is now generally available
https://techcrunch.com/2023/06/07/googles-generative-ai-support-in-vertex-ai-is-now-generally-available/

If you’re sick of AI chatbots I have bad news for you and your Instagram DMs
https://www.techradar.com/news/if-youre-sick-of-ai-chatbots-i-have-bad-news-for-you-and-your-instagram-dms

Blush, the AI lover from the same team as Replika, is more than just a sexbot
https://techcrunch.com/2023/06/07/blush-ai-dating-sim-replika-sexbot/

Automattic launches an AI writing assistant for WordPress
https://techcrunch.com/2023/06/07/automattic-launches-an-ai-writing-assistant-for-wordpress/

OpenAI Sued for Libel After ChatGPT Allegedly Hallucinates Man Into Embezzlement Lawsuit
https://gizmodo.com/chatgpt-openai-libel-suit-hallucinate-mark-walters-ai-1850512647

Benchmark’s view on the AI race: talking with Miles Grimshaw
https://techcrunch.com/2023/06/06/benchmarks-view-on-the-ai-race-talking-with-miles-grimshaw/

Microsoft has no shame: Bing spit on my ‘Chrome’ search with a fake AI answer
https://www.theverge.com/2023/6/6/23736289/microsoft-bing-chrome-search-fake-ai-chatbot

Edtech giant Byju’s launches transformer models in AI push
https://techcrunch.com/2023/06/06/edtech-giant-byjus-launches-transformer-models-in-ai-push/

OpenAI CEO Sam Altman is privately reassuring developers using the company's tech that it won't compete with them beyond ChatGPT
https://www.businessinsider.com/openai-sam-altman-software-developers-ai-build-consumer-products-chatgpt-2023-6

OpenAI still not training GPT-5, Sam Altman says
https://techcrunch.com/2023/06/07/openai-gpt5-sam-altman

Contextual AI launches from stealth to build enterprise-focused language models
https://techcrunch.com/2023/06/07/contextual-ai-launches-from-stealth-to-build-enterprise-focused-language-models/

DeepMind repurposes game-playing AIs to optimize code and infrastructure
https://techcrunch.com/2023/06/07/deepmind-repurposes-game-playing-ais-to-optimize-code-and-infrastructure/

You can now preorder Google's new AI tool for Workspace
https://www.techradar.com/news/you-can-now-preorder-googles-new-ai-tool-for-workspace

Black Mirror Creator Says He Used ChatGPT to Write an Episode. It Was Terrible.
https://gizmodo.com/black-mirror-season-6-chatgpt-writers-strike-ai-1850513967

FBI Warns of AI-Generated Deepfakes Images in Sextortion Schemes
https://petapixel.com/2023/06/07/fbi-warns-of-ai-generated-deepfakes-images-in-sextortion-schemes/

Google claims that Bard is improving at math and programming
https://techcrunch.com/2023/06/07/google-claims-that-bard-is-improving-at-math-and-programming/

#AI #ArtificialIntelligence #ML #MachineLearning #LLM #ChatGPT #BingChat #SwarmsOfRampantAIChatbots

Note they claim a 30% better response.

#Google updates Bard with improvements to mathematical tasks, coding questions, and string manipulation via a technique that lets Bard run code in the background.

https://techcrunch.com/2023/06/07/google-claims-that-bard-is-improving-at-math-and-programming/

#AI #ML #MachineLearning #GenerativeAI #LLMs

#будущее #нейропиздобол #ml @rf @ru
Подвезли рейтинг доступных для скачивания нейропиздоболов: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

Скачал Wizard-Vicuna-13B-Uncensored-HF как наиболее пригодную для работы на проце хинкпада - работает и правда ощутимо лучше всяких лламо-альпак при том же потреблении ресурсов (llama.cpp, AVX2, q4_0 квантизация). Даже стишки про ниггеров пишет с переменным успехом. Но польза всё ещё околонулевая: нормального кода из неё сгенерить не вышло, а тем временем датацентровые кластерные проприетарщики бороздят просторы вселенной: https://chat.openai.com/share/c24445cd-4c82-4a72-bbd0-37ebcc6135f7

Stumbled upon a Reddit thread where someone trained an img-to-img #ML model to take QR Codes and embed them into generated imagery while keeping the codes fully functional! The results are stunning.

8 images total available in the original thread:

https://www.reddit.com/r/StableDiffusion/comments/141hg9x/controlnet_for_qr_code/

🔖 The Top 5 Papers About #mlops You Should Know (Part 2/4)

Not just another list of #ml, #datascience, #deeplearning papers.

1️⃣ Operationalizing Machine Learning: An Interview Study By
Shreya Shankar, Rolando Garcia, Joseph M. Hellerstein, Aditya G. Parameswaran (https://arxiv.org/abs/2209.09125 )

2️⃣ Socio-Technical Anti-Patterns in Building ML-Enabled Software by Alina Mailach, Nortbert Siegmund (https://sws.informatik.uni-leipzig.de/wp-content/uploads/2023/01/socio-technical-anti-patterns-icse2023.pdf )

Calling all #Minecraft fans! 🎉

Meet Voyager - an #LLM power agent that utilizes #GPT4 to engage in Minecraft gameplay. It demonstrates remarkable capabilities by learning, retaining knowledge & showcasing exceptional expertise in Minecraft.

Read more on #InfoQ and get ready for an unforgettable gaming experience: https://bit.ly/43JpgyP

#AI #ML

A lovely deep dive 70+ page paper about neural network architectures and embeddings in large language models, a personal project by Vicki Boykis https://jawns.club/@vicki - thank you, Vicki!

personal intro: https://vicki.substack.com/p/what-are-embeddings

site: https://vickiboykis.com/what_are_embeddings

pdf: https://raw.githubusercontent.com/veekaybee/what_are_embeddings/main/embeddings.pdf

code: https://github.com/veekaybee/what_are_embeddings

fedi: https://jawns.club/@vicki/110492295059859119

#ml #NeuralNetworks #LLM

If you think Apple is not interested in what everyone is calling #AI, just look at the #WWDC sessions on machine learning. They've had these sessions for years. While Microsoft, Google, and Amazon try to sell their cloud infrastructure for #ML use, Apple's already got their platform in play and has done so for those multiple years.

#GenerativeAI #MachineLearning #GPT

I'd say for better.

For better or worse, #Apple is avoiding the #AI hype train / At this year’s WWDC, the iPhone maker shunned references to ‘AI’ in favor of the more sedate ‘machine learning.’ But for how long will this caution be beneficial?

https://www.theverge.com/23749697/apple-wwdc-ai-machine-learning-new-features-chatbots-chatgpt

#MachineLearning #GenerativeAI #ML

The Power of Linux. Zero Setup. No Kidding.
See our systems at: https://kfocus.org/

#Linux #Laptop #Desktop #ML #Cybersecurity #Tech #AI #DeveloperTools #Creators

I find it interesting to watch the look on folks faces as I explain that using #ML / #AI coding tools (#copilot, #android #studiobot, etc.) is like giving access to your code to an unhelpful coder;

It won't sign a company specific employment contract, NDA, etc., and so it has little accountability for what it gives you.

It won’t help you learn about the problem domain, or what trade-offs and alternatives it considered.

It won't mentor you to be come a better programmer.

It is a tool.

Kind-of-fun opinion on one ACM blog: "AI Does Not Help Programmers" https://m-cacm.acm.org/blogs/blog-cacm/273577-ai-does-not-help-programmers/fulltext
However, there's a (side note) sentence that caught my eye: «I am talking about "modern" AI to distinguish it from the earlier kind—largely considered to have failed—which tried to reproduce human logical thinking».
One ear that a lot since the raise of ML, but is it is true? Did "classical AI" failed to give anything useful around the coding assistance discussed in the blog post?
Of course not.
#AI #ML

LLMs as coders? “the resulting programs don't actually do what they were asked to do“ https://languagelog.ldc.upenn.edu/nll/?p=58977 #ml #ai

If you have a client that needs thousands of product categories remapped, what do you do?

Well, if you're a programmer you'll run it through a huggingface text classifier, of course! 😆

#tech #programming #software #ai #ml

Do you like big data? So does the Kubuntu Focus!

"A TensorFlow job took 37 hours on my MBP. It took just 1.5 hours on my Focus."

Visit us and find out more: https://kfocus.org/

#ML #AI #BigData #Robotics #Linux #DataScience #Engineers

Spending the morning looking at the impact of hard attack in online communications and #NLP #ml

#AI is eating Andreessen

– or –

Andreessen is eating #AI

🙄

#BigTech
#ML #BigData
#SiliconValley
#TechnoFascism
#Technofascist
#VC #AIBlitzkrieg
#OligarchBlitzkrieg
#Surveillance
#SurveillanceCapitalism
#Privacy
#Democracy
#NatSec
#USpoli

Announcing the launch of new @huggingface LLM Inference containers on Amazon SageMaker 👉 Powered by Text Generation Inference (TGI), an #OpenSource, purpose-built solution for deploying and serving Large Language Models https://aws.amazon.com/blogs/machine-learning/announcing-the-launch-of-new-hugging-face-llm-inference-containers-on-amazon-sagemaker/ #AWS #MachineLearning #ML #GenAI

W końcu ML przydał się do czegoś sensownego: kody QR, które nie wyglądają jak kody QR 🤩

https://www.reddit.com/r/StableDiffusion/comments/141hg9x/controlnet_for_qr_code/

#ML #AI

Know what's been baked into majority of #Apple HARDWARE DEVICES for years and years and years now?

– a little something called Machine Learning #ML

*plus*

– metric crap-tons of something #TimCook barely mentioned (if at all) yesterday
👇
#AI
#ArtificialIntelligence
#Privacy
#WWDC23
#WWDC
#WWDC2023
#VisionPro
#AppleVisionPro

According to an interview on RNZ, the NZ DIA (Department of Internal Affairs)...

"... are formulating some advice on how the public service would be able to utilise the benefits of AI..."
https://www.rnz.co.nz/national/programmes/focusonpolitics/audio/2018892689/politicians-tackling-ai-regulation-but-will-it-be-too-late

Anyone know how we feed into that?

#RNZ #podcasts #FocusOnPolitics #AI #ML #MOLE #Aotearoa #NZ #DIA

😴 Tired: Yet another list of 10 AI tools that 1000x your life

🔥 Wired: A list of the 10 overhyped AI tools that failed miserably

🎯 Challenge: Can anyone create that list to save us?

🫡 I promise I'll boost it 🚀

#AI #ML #chatgpt #mastodon #productivity

😴 Tired: Yet another list of 10 AI tools that 1000x your productivity

🔥 Wired: A list of the 10 overhyped AI tools that failed miserably

🎯 Challenge: Can anyone create that list to save us from endless productivity?

🫡 I promise I'll boost it 🚀

#AI #ML #productivity #hype

An AI team at Nvidia created a bot that uses GPT-4 to solve problems inside Minecraft.

https://www.wired.com/story/fast-forward-gpt-4-minecraft-chatgpt (4-min read)

#AI #ML #GPT4 #nvidia #tech #technology #minecraft #automation

I was pretty disappointed with @caseynewton and Kevin Roose's interview with Ajeya Cotra on Hard Fork.

Cotra seems to misunderstand some really fundamental things about machine learning and neural networks, two of which in particular lead to her entertaining bizarre and implausible "doomsday" scenarios:

She repeatedly claims that for LLMs to accurately do next-token prediction, they must have an "understanding" of the topic. ("In order to predict what comes next in a science textbook after reading one paragraph, you have to understand something about science.") Charitably, my read of the literature is that this is a controversial opinion at best; I think @emilymbender 's "Thai Library" thought experiment does a good job illustrating why this doesn't really make sense for any meaningful definition of "understanding."
She doesn't seem to know how model training actually works? This is a bit subtler, but her doomsday scenario involves a model gaming its own feedback ("you can think of it as like pushing the reward button when it does well and pushing the anti-reward button when it does poorly...the best way to do so may not forever be to just sort of basically do what the humans want...It might be something more gain access at a root level to the servers that Google is running and, with that access, be able to set your own reward."). But in modern neural networks, models are trained and then weights are frozen before the model is evaluated.

Thus, the model doesn't "want" to get the "reward button" (or as Cotra says, "setting the rewards that are coming in to be high numbers"). During evaluation, it doesn't even know about the "reward button"! The feedback is only used during training to optimize model weights. So while a bad objective function might lead to a trained model that does bad things, the simplistic story of an LLM that "wants" a reward and, like a drug addict, violates its own training to get it just doesn't translate to how NNs actually work. It just doesn't make sense to say the model will "set the rewards to high numbers", unless, during training, it was given the ability to set the rewards.

These two fundamental errors lead to a doomsday scenario in which a model is creative enough to behave differently (and deceitfully) in production than it had during training (because it has "understanding"), and in which it "wants" its "reward" function and is willing to go to desperate lengths to get it.

But this just doesn't resemble how any modern ML works! The whole discussion is basically fan fiction!

(I'm trying something new here and deliberately replacing every place where I'd normally write "#AI" with "#ML". It seems kind of illuminating!)

Push #ML to the #Edge in #IoT personal information is kept at wearables/devices collecting #health data, aggregating, orchestrating and analyzing data where needed throughout sensor analytics ecosystems #SensAE #P4Health

“To help with early detection, researchers in the UK are proposing a phalanx of #iot devices, #ai , and #cloudcomputing to provide real-time alerts when someone is experiencing a suspected cardiovascular emergency.” — @czds

https://spectrum.ieee.org/heart-disease-iot-detector

https://mastodon.social/@czds/110482949859930501

Tim Hickson aka "Hello Future Me" - a former IP attorney and current writer and video essayist - just dropped a feature-length video on the issues around AI art. It's available on both YouTube and Nebula.

This one is as important as Line Goes Up; you should give it a watch (we're about 45 minutes in and saving the rest for a drive later today; so far my attorney wife is very satisfied with the analysis):

https://youtu.be/9xJCzKdPyCo

#AIArt #ML

‘Hey Alexa... get the f*ck out of my house”

👇

“How voice data is being used with AI to generate "predictive" medical diagnoses” | Boing Boing https://boingboing.net/2023/06/03/how-voice-data-is-being-used-with-ai-to-generate-predictive-medical-diagnoses.html

#AI #ML
#MachineLearning
#ArtificialIntelligence
#Audio
#Surveillance
#BioSurveillance
#Privacy
#Health
#HealthTech
#DigitalHealth
#DigitalMentalHealth
#SurveillanceCapitalism
#MedMastodon

Processing a video (people walking at Time Square) with #computervision on m1 running with #asahi Linux. :asahi: #AI #Yolov5 #pytorch #ml @AsahiLinux @lina

Come and apply your #GenerativeAI #MachineLearning skills to social good at Nesta! https://www.nesta.org.uk/procurement-and-invitations-tender/call-for-an-expert-to-harness-generative-ai-for-social-good/

#ML #AI #GPT #TechForGood #Jobs #LanguageModels #LargeLanguageModels #LLM #ChatGPT #Intelligence #ArtificialIntelligence #GPT4

Attacks Against Apache NiFi #apache #nifi #ml #cryptominer #bot https://i5c.us/d29900

I have created a "Medical AI" list of mastodon users to help colleagues get started more quickly #twitterMigration

You can add yourself here:
https://followlists.online/list/42dd6cc8-e347-d0b1-fcdc-72e9c254626c

#MedicalAI #ML #CDSS #CDS #eHealth #DigitalHealth #med #AI

💥 Exciting updates in the #AI, #ML & #DataEngineering world:
➡️ Adobe has introduced a new feature called Generative Fill in Photoshop
➡️ Microsoft is bringing the power of AI to Windows 11
➡️ Opera reveals Aria, the integrated AI browser feature
➡️ Swiss researchers utilize AI to reconstruct spinal cord

Read on #InfoQ: https://bit.ly/3MEUmkh

concerning recent papers in #VoiceConversion #ml #research #science #audio #speech

A group of prominent #AI and #ML scientists signed a very simple statement on giving the possibilities of global catastrophe caused by AI more prominence.

https://www.safe.ai/statement-on-ai-risk

This is part of a broader movement of #AISafety or #AIRisk. I don't disagree with everything this movement has to say; there are real, and tangible consequences to unfettered development of AI systems.

But the focus of this work is on possible futures. Right now, currently, there are people who experience discrimination, poorer outcomes, impeded life chances, and real, material harms because of the technologies we have in place *right now*.

And I wonder if this focus on possible futures is because the people warning about them *don't* feel the real and material harms #AI already causes? Because they're predominantly male-identifying. Or white. Or socio-economically advantaged. Or well educated. Or articulate. Or powerful. Or intersectionally, many of these qualities.

It's hard to worry about a possible future when you're living a life of a thousand machine learning-triggered paper cuts in the one that exists already.

It seems like we have entered the "Letters of AI" stage.

Letters, letters everywhere, nor any words to think.

#AI #ML #responsibleAI #mastodon #ethics

Today's the Day! 🥂

We're hosting a workshop to get folks started with #RLHF!

Join Jimmy Whitaker, Nikolai Liubimov, and myself for an entry-level workshop on Reinforcement Learning with Human Feedback.

📅 May 30 — 2-3 PM EDT

🔗 https://lu.ma/RLHF

#ReinforcementLearning #Tutorial #MachineLearning #ML

Why do all these Bostromians not see the obvious solution to their anxiety: No #AI ?

Because that might threaten *their* jobs?

And why do they consistently forget to add the #climatecrisis to their list of extinction risks?

#llm #ml #ethics #politics #ideology #bostrom #longtermism #capitalism

https://www.safe.ai/statement-on-ai-risk

Join the 4th Workshop on #UserCentered #AI (#UCAI2023) at @mucConf in Lake Zurich,🇨🇭

Submission Deadline: June 16th, 2023 
Notification: July 3rd, 2023
Workshop: September 3rd, 2023

CfP: https://ucai-sig.org/events/ucai23/call/

#UCAI23 #MuC2023 #ML #CHI2023 #IUI2023 #UIST2023 #XAI

PDFs per ML taggen lassen – Adobe hat dafür eine durch maschinelles Lernen gestützte API angekündigt: „Adobe PDF Accessibility Auto-Tag API“.
Getaggtes PDF ist notwendig, um den barrierefreien Zugang zu den Inhalten von PDF-Dateien zu ermöglichen. Dies nachträglich händisch durchzuführen, ist mühsam.
Anwender, die eine Adobe-ID haben, können die Funktion testen. Als Download erhält man ein PDF/UA und eine Log-Datei.

#PDF #Tagging #Accessibility #AdobeSensei #ML #KI

https://acrobatservices.adobe.com/dc-accessibility-playground/main.html#

Lots of Prompt Engineering classes out there & relatively little talk about the dangers of inserting untrusted data into LLM prompts.


Feels like the 90s and everyone builds vulnerable systems at scales.

https://www.youtube.com/watch?v=Fz4un08Ehe8

#redteam #aiinjection #LLM #PenTest #openai #ml #langchain #infosec

Case in point:

https://www.engadget.com/a-lawyer-faces-sanctions-after-he-used-chatgpt-to-write-a-brief-riddled-with-fake-citations-175720636.html

#ChatGPT #AI #ML

Tech companies invest millions on their technical infrastructure.

Very little is invested in building their sociotechnical infrastructure.

This is sad because the problems they tackle are sociotechnical in nature.

Technical solutions don't work for sociotechnical problems.

#AI #ML #technology #sociotechnical

@AmyDentata @mattedgar

What #Chatbots do extremely well?

•  harvest human interaction #data and associate #PersonalData to train and improve their own algorithmic backend systems

#AI #ML
#BigData
#Privacy
#SiliconValley
#BigTech
#OpenAI
#ChatGPT
#TechnoFascism
#OligarchBlitzkrieg
#AIBlitzkrieg
#ElonMusk
#PeterThiel
#SamAltman
#AIEthics
#EthicalAI
#Democracy

You'd be statistically wise to do the opposite of anything I advise, stock-wise, but: Nvidia's lock on the #ML market is as much a matter of software (CUDA) as hardware. And it seems to me that the CUDA wall has cracks (not crumbling!). I don't know if we're at peak #NVDA, but I don't think they have a lot of headroom left.

Explainable AI peeps, do you know when "open the black box" narrative was stated for the first time?

I'm trying to hunt down a historical citation. It doesn't have to be a peer reviewed piece as long as it was the first time that phrase was used in the context of XAI

w/ @Riedl
#AI #ML #XAI

@ricmac
Steve Albini @electricalWSOP Musician & Recording Engineer, succinctly calls the generated music hype BS:

» #AI is predicated on notion that #ML can learn what I like or expect, and give me that..

baffling to me: things that are unexpected!

things once rare & difficult to obtain.. become ubiquitous..

#musictech danger is not the power.. but natural flow..
argues for banality, conformity, and the mundane..
against the unexpected, accidental, perverse «

~1:25
https://youtu.be/NZgmXaUYeKE?t=5100

New trend in #ML:
Don't have training data?

Just generate it using #chatGPT/#LLM!

Aite, cool.

But you do realize using synthetic training data and then moving to real-world deployment ain't cool, right?

Do the ✨hard work✨of non-extractive data collection 🫶

#AI #NLP

on the occasion of #TowelDay ,
watching a remake -
and these words now sound oddly prophetic:

Ford #Prefect: [about #Vogons] "They don't think, they don't imagine, most of them can't even spell. They just run things."

#AI #ML #LLM #thereIsNoAI #chatgpt

Feels like combining #llm models with what Semantic Web is doing for a long time with triple stores and ontology modelling would be a good idea to enhance #ml #machinelearing #artificialinteligence #ai algorithms, has that been done already?

My dream comes true that this story finally gets told...

Thank you #Apple
🙏

#Health #DigitalHealth #HealthTech #MentalHealth #DigitalMentalHealth #AI #ML
#HealthKit #CareKit #BigData #HealthAI #AIEthics #EthicalAI #Privacy
#ArtificialIntelligence
#MachineLearning
#SiliconValley #BigTech
#PersonalData #HealthData
#Surveillance #BioSurveillance
#SurveillanceCapitalism
#SmartPhones #Wearables
#MedMastodon #AppleWatch

90 seconds:

#Business #Launches
Generative AI: what you need to know · A guide to learn the basics of generative AI and to spot AI bullshit https://ilo.im/135281

“It’s a crash course in AI snake oil.” — Baldur Bjarnason

_____
#AI #ML #LanguageModel #Pitfalls #Dangers #Design #ProductDesign #UxDesign #UiDesign #WebDesign #Development #WebDevelopment #WebDev #Code #Safety #Privacy #Copyright #Plagiarism #Poisoning #Fraud #Abuse #Bias

Starting with the mechanics of understaning how LLMs work can help foster durable intuitions that will inform our usage of these models now & in the future. (Especially if the future is one where LLMs are a staple of the data scientist’s toolbox, as common as an lm() function call).

And what better way is there to learn than by doing!

Tomasz Kalinowski walks through an implementation of LLaMA, a Large Language Model, in R, with TensorFlow and Keras.

https://blogs.rstudio.com/ai/posts/2023-05-25-llama-tensorflow-keras/

#rstats #ai #ml

@blakereid
As long as it has #AI #ML and #BigData – each iteration will taste better 😉

💡 A Build 2023 attendee posted this interesting US copyright link in a session's chat today...

Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence
https://www.federalregister.gov/documents/2023/03/16/2023-05321/copyright-registration-guidance-works-containing-material-generated-by-artificial-intelligence

👆 The link above is better to use than the auto-generated link below 👇

#Microsoft #Build #Build2023 #MicrosoftBuild #Cloud #Mesh #ChatGPT #GPT #OpenAI #Edge #Bing #GitHub #Azure #AI #ML #AR #VR #MR #XR #Windows #Accessibility #Inclusion #TrainingData #Copyright #Infringement

This piece around "superintelligence governance" has sparked quite the uproar.

But is it all just hype?

Let's take a closer look at why some believe it's much ado about nothing.

First and foremost, there seems to be a fog of confusion surrounding the very definition of "superintelligence." Even if we consider Nick Bostrom's interpretation (or lack thereof), the way OpenAI employs the term leaves us scratching our heads.

https://openai.com/blog/governance-of-superintelligence

#ML #AI #OpenAI #responsibleai

1/n

💡 This Build 2023 #ComputerVision session was interesting - Using AI #VectorSearch to locate / identify items in images and frames of video

"Build Apps with State-of-the-Art Computer Vision" 👀
https://build.microsoft.com/en-US/sessions/6192a166-2742-42c1-9d12-adbb3daa30bb

Azure AI Vision Studio
https://portal.vision.cognitive.azure.com/gallery/featured

Registration
https://register.build.microsoft.com

#Microsoft #Build #Build2023 #MicrosoftBuild #Cloud #Mesh #ChatGPT #GPT #OpenAI #Edge #Bing #GitHub #Azure #AI #ML #AR #VR #MR #XR #Windows #Accessibility #Inclusion

GPT - Generatively Pre-trained Transformer

"Let's Build GPT. From Scratch. In Code. Spelled Out" by Andrej Karpathy
https://www.youtube.com/watch?v=kCc8FmEb1nY
(Coding starts at 42:13 mark)

"ChatGPT... is trained on a good chunk of the Internet" (3:26 mark) 🤔 <-- Meet Mr. Copyright ©

Andrej Karpathy
https://build.microsoft.com/en-US/speakers/1db4b2b5-c5df-4372-aaae-c96b6fa93ef8?source=/speakers/1db4b2b5-c5df-4372-aaae-c96b6fa93ef8

"Attention Is All You Need" Seminal Paper on AI Transformers
https://arxiv.org/abs/1706.03762

#Microsoft #MicrosoftBuild #Build2023 #ChatGPT #GPT #OpenAI #AI #ML #Copyright #Infringement

🚩 The "State of GPT" session at Microsoft Build 2023 was loaded with great info (fast-paced compressed presentation needs additional viewing)

"State of GPT"
https://build.microsoft.com/en-US/sessions/db3f4859-cd30-4445-a0cd-553c3304f8e2?source=sessions

Registration
https://register.build.microsoft.com

#Microsoft #MicrosoftBuild #Build2023 #ChatGPT #GPT #OpenAI #AI #ML

While still in the research phase, this ultrasound system on a patch #USoP is an important advancement for deep-tissue cardiac monitoring, both at a medical facility, and with machine learning #ML adjustments for movement, while going about one’s normal activities. #Wearables #IoMT #InternetOfThings #IoT bring #TeleMedicine #mHealth to underserved populations around the globe to make #healthcare Sensor Analytics Ecosystems #SensAE more broadly available.

https://www.thebrighterside.news/post/scientists-develop-the-first-fully-integrated-wearable-system-for-life-saving-heart-monitoring

Regarding my Microsoft Build 2023 posts, I just drank the Cool-Aid® 🥤 😗

The Digital Access Pass is Free

Registration
https://register.build.microsoft.com

#Microsoft #Build #Build2023 #MicrosoftBuild #Cloud #Mesh #ChatGPT #GPT #OpenAI #Edge #Bing #GitHub #Azure #AI #ML #AR #VR #MR #XR #Windows

where are my not-men in ML at? I feel like every event I go to is a giant sausage fest. 🥲

#WomenInStem #MachineLearning #ML #OpenSource

“Calling it ‘content’ was the first step in convincing people that machines could make it.” — Peter Ramsey

_____
#Design #AI #ML #Chatbot #ProductDesign #UxDesign #DigitalDesign #WebDesign #WebDevelopment #Website #Blog #Content #Human #Machine #Quote

OK my new podcast is called #DataRevolution, first episode is live https://feeds.buzzsprout.com/2192536.rss

#AI #ML #data

Take a technical deep dive with @AMD CVP Raghu Nambiar as he breaks down how 4th Gen AMD #EPYC processors deliver generational performance uplifts on several #AI and #ML workloads, including ResNet-50, BERT-Large, and DLRM.

[Re. @AMD]
NOW AVAILABLE: The TensorFlow-ZenDNN plug-in for @TensorFlow v2.12 and above. This enables neural network inferencing on @AMD #EPYC CPUs and improves the accessibility of ZenDNN optimizations via framework upstreaming. https://t.co/5En89YgdWG #NLP #ML #AI

[Re. @AMD]
AMD is committed to support open source software and are excited to partner with @Google on #OpenXLA. Learn how we are working with OpenXLA to support cross platform portability for #ML models across CPUs, GPUs, FPGAs and more. https://t.co/b6iYpqT7of

Read about @iitjodhpur deploying @Lenovo #ThinkSystem servers powered by @AMD #EPYC to provide more compute performance than its previous #HPC cluster, future-proofing its infrastructure for intensive scientific workloads including #AI and #ML. https://bit.ly/3E5Vn16

🎥 Tune in now to learn how to use GitHub to make Machile Learning education more accessible and extendable: https://bit.ly/3Eswqhj
@github #GitHubUniverse #GitHubeducation #ML #AI

[Re. @AMD]
"Together with other foundation members, we will support the acceleration of science and research that can make a dramatic impact on the world." - AMD Data Center and Accelerated Processing CVP Brad McCredie #AI #ML #HPC #PyTorch https://t.co/E1WvtwbjRu