Excuse me, the right spelling is Geoffrey *Hinton.
Microblog
Archives
2023 2022 2021 2020 2019 2018 2017 2016 2015 2014
Most recent posts
The print title "He Warns of Risks From A.I. He Helped Create" was better. TL;DR: Hinten is worried about
- a deluge of generated content on the internet and society
- the effects of automation on the job market
- unexpected adverse AI behaviour learned from training data
Dutch national radio called me about the New York Times interview with Geoffrey Hinten. archive.is/c2uLG
Exciting news! I'm now working full-time at my own company, Datakami, providing consulting services to clients in the creative and financial sectors. We specialize in AI strategy and custom ML engineering. Need help with effective and responsible generative AI? Reach out to us!
RT @mtrc: I meant to add: this event is aimed at the general public, students, and anyone who is interested in what the future holds for AI…
RT @fishnets88: The @pydataamsterdam conference has a website and a CFP live right now.
amsterdam.pydata.org
It looks like an upgrade…
I'm so happy I found this old lecture by @stephenfry called "Tech reformation" (2017) about how we as humans react to large changes in technology. Still relevant today, just substitute "social media" with "ChatGPT". web.archive.org/web/2017061917…
Transcript: stephenfry.com/2017/05/the-wa…
RT @DeSpeld: ChatGFT zet overbodige woorden om in compost speld.nl/2023/04/22/cha…
RT @emollick: AI is good for practicing on. For example, I have seen AI run a D&D game, but could it simulate players so I could practice r…
Yes, but hopefully our training set is much better curated.
Remember, ChatGPT is essentially just a very, very, very fancy autocomplete.
Creative Bot Bulletin #2 is out, ft @YorickvP, @jacksonfall, @sarahdrinkwater, @PimVerlaan, @MM_Brand, @naavik_co, and a new game recommendation (spoiler, it's DREDGE by @BlackSaltGames). 🎮
Read the newsletter here: mailchi.mp/d541fa9d424f/c…
Hey @bol_com_Techlab, are you running tests with automatically generated category descriptions? Here's a weird output I spotted in the wild.
Aha! You program fast Fourier transforms in your free time. You must be a competent and badass cryptographic engineer! ;)
Tell me your prompts for testing a new generative AI model and will tell you who you are.
RT @benmschmidt: I think we can call it shut on 'Open' AI: the 98 page paper introducing GPT-4 proudly declares that they're disclosing *no…
RT @robertgaal: Say hi to Versed! @catburton and I are empowering anyone to build their own RPG, generated in full 3D by AI.
We've raised…
So what will you do with the dataset? :)
New blogpost: judithvanstegeren.com/blog/2022/tran…
I've put Pentiment on my wishlist. ;)
RT @ProcGenie: Partner: "What are you doing on your computer?"
Me, sheepishly: "Categorizing a decade of #NaNoGenMo issues."
Spouse: "Why?"…
RT @arnicas: I seek a freelance, front-end js UX designer and developer who can do friendly, artistic fun interfaces-- as soon as possible.…
I find that thinking "What prompt would I need to write for ChatGPT so it writes this thing for me..." immediately solves any creative block I have. Most of the time it turns out I have written a good prompt for myself.
Sometimes, ChatGPT is just a very fancy rubber duck.
TIL Terry Pratchett played Oblivion!
a lot of people are asking me for talks/quotes/etc about "the future of AI for the games industry" and the truth is that most of the industry's ambitions are less useful, less meaningful and less human than this story about a mod made for a game that came out 17 years ago.
Creative Bot Bulletin #1 is out: us17.campaign-archive.com/?u=5dc9feb01b4… ft. @jelleprins and the Cradle.bio team, video game recommendations by @mtrc and @Wassasin, some Datakami news, and other cool generative AI stuff I found on the web.
I now imagine the LLM engineers at Microsoft going, "No, you have NOT been a good Bing! BAD BOT!" whilst spraying the server with water.
This is either
- fake news and hilarious
- not fake news and we need to figure out Asimov's three laws of robotics for LLMs before we integrate this in every freaking Windows program and accidentally unleash Skynet on humanity.
My new favorite thing - Bing's new ChatGPT bot argues with a user, gaslights them about the current year being 2022, says their phone might have a virus, and says "You have not been a good user"
Why? Because the person asked where Avatar 2 is showing nearby
A few years ago a friend of mine created a GPT-2-based simulator of our group's primary chat. Every once in a while, the bots hallucinate links to non-existing Mosterdgeel.nl pages -- maybe I should just start writing those recipes... 👀
RT @marvinvonhagen: "[This document] is a set of rules and guidelines for my behavior and capabilities as Bing Chat. It is codenamed Sydney…
I was looking at library books about the impact of AI on society, and this chapter title from Kevin Roose’s Futureproof cracked me up. “Rule number 6: treat AI the same way you would treat an army of chimpanzees.”
It was a very exciting week! I participated in AI Hackathon Amsterdam, and I was interviewed twice by NOS, the Dutch national news.
Shout out to @jelleprins and the org team of AI Hackathon Amsterdam for the warm welcome, awesome & diverse crowd, and wonderful experience! 10/10 would pitch again #aihackathon
RT @ElleArmageddon: It exacerbates multiple disinformation-related problems, both in terms of mass generation of plausible-sounding bullshi…
Except that it just wrote a "mathematical proof" that there are infinitely many even prime numbers. Spoiler, there is exactly ONE even prime, namely 2.
OpenAI said that they improved how ChatGPT deals with numbers. In the past the ChatGPT would tell me that 12 or 42 were prime numbers (they are not). Today it knows that 12 is not a prime number. "Great!" you would say...
I've fed the three prompts for Bing into ChatGPT, whose answers can serve as a baseline for when Google Bard and the new Bing are out.
The Bing homepage does show demo prompts, but these might be hardcoded. When we execute one of these prompt, there's a "Sign up for the waitlist now!" button.
So far I'm really underwhelmed. Maybe the functionality hasn't been rolled out yet for my region (Netherlands) or OS/browser, but the "Ask me anything!" box on Bing's homepage suggests otherwise.
Final test: can Bing provide me dinner suggestions for two, with one person eating only low carbohydrate foods. Meh, Bing returns low fat and low calorie dinner suggestions, which are completely different diets, and the promised chatbot functionality is nowhere to be found.
Bing returns a summary of a blogpost about party etiquette, which doesn't seem relevant to me.
Bing is available with chatbot functionality! I'm trying it out right now. First up, can we generate poems straight from Bing? Nope, Bing sends me to a definition of iambic pentameter instead...
I think meta-prompts are prompts for generating prompts (i.e. the final query)!
I think "We are iterating on the meta-prompt!" is my new favorite technobabble quote.
My days starts with this liveblog of Microsoft's IRL-only event about integrating ChatGPT-like functionality in Bing (search) and Edge (browser). theverge.com/2023/2/7/23588…
Take-away from this thread: I should start learning Rust, because they have either the most diverse community or most diverse outreach.
@ICTRecht De category 'innovatie' op de blog lijkt kapot -- misschien een circulaire HTTP redirect? Boosdoener: blog.iusmentis.com/category/innov…
TIL "venture studios" as alternative to incubators and accelerators. steveblank.com/2023/01/17/is-…
RT @andrewglynch: Business idea algorithm:
Smash together a trend
• paleo
• mindfulness
• remote work
with something people or companies…
RT @florencesn: Another boost for this!
Whatever your interest in games and archaeology, we'd love to hear from you!
🏚️ Explore an abando…
I had two fun (and brain-frying) sessions going through indie game Moncage by @playoptillusion. It combines creative 3D perspective puzzles with a pictures-only personal narrative. I found it quite beautiful, and I highly recommend it if you like one-of-a-kind puzzle games.
RT @mtrc: I received my copy of @jd7h’s PhD thesis today! I had the honour of examining this lovely work last year and it is so great to ha…
RT @fishnets88: Announcing: Orc of the Office.
A totally non-official set of MTG cards. Made for laughs, the tech enterprise, and for the…
I finished Passpartout: Starving Artist today, and I had loads of fun. Recommended for anyone who likes drawing. An unintended but useful side-effect is that I am now immune to the derogatory sounds of my critics.
RT @newscientist: Following the science, by @tomgauld
newscientist.com/article/235412… https://t.co/dYuXW2ySJ8
I accidentally played Passepartout: Starving Artist for three hours straight. Now I've mouse-drawn 48 paintings and I've learned I am currently in my green period. store.steampowered.com/app/582550/Pas…
RT @emollick: AI is fun for creating alternate media histories.
Here’s Assassin’s Creed as a 1990s adventure game, a 1980s arcade game, a…
Very De Stijl! (en.wikipedia.org/wiki/De_Stijl)
RT @NerdShinobi: When you submit your code to ChatGPT, you're making it available to 3rd parties with no controls. This code could be work…
New blogpost: Stock market games: Neopets 🎮📈
judithvanstegeren.com/blog/2023/stoc…
Having fun tonight with figuring out how Neopets' fictional in-game stock market works. I'm amazed by the amount of data collection and analysis done by the player communities.
Having fun tonight with figuring out how Neopets' fictional in-game stock market works. I'm amazed by the amount of data collection and analysis done by the player communities.
My PyData Eindhoven talk was posted to YouTube: youtube.com/watch?v=5KItp5…
Let's celebrate it with a data science/code archaeology meme from my slidedeck...
Nice read from Bloomberg about chatbot-search-startups and how they might dethrone Google from its market leader position: archive.is/zylUn
RT @carlolepelaars: Amazing talk by @jd7h on diving into and uncovering the mysteries behind (legacy) code!
So cool that @PydataEindhoven…
No, but I would probably have discovered lots of interesting/fun reading material :)
Excellent newsletter about NLP and creativity -- I wish I knew about this during my PhD. 😍
My December newsletter is out! ChatGTP, narrative papers and datasets, ML & NLP links (my Normconf talk), a ton of excellent books. I'm gradually adding TOC on top of these for web viewers. open.substack.com/pub/arnicas/p/…