- That AI Thing
- Posts
- weekend ai reads for 2024-11-08
weekend ai reads for 2024-11-08
📰 ABOVE THE FOLD: AUTHENTICATION
AI Can Now Authenticate Sneakers by Their Smell — An AI start-up working with a major sneaker resale platform believes it has found a way to authenticate footwear and potentially other products through the chemical signatures in their scent. / Business of Fashion (8 minute read)
Meta Plans to Use AI to Catch Teens Lying About Age on Instagram / Bloomberg (7 minute read)
Aesthetic launches the ‘Shazam for fashion’ / Tech Crunch (6 minute read)
Beyond Passwords: AI’s Role in the Future of Authentication / Security Boulevard (9 minute read)
📻 QUOTE OF THE WEEK
If you’re thinking without writing, you only think you’re thinking.
Leslie Lamport (source)
👥 FOR EVERYONE
What Trump’s victory could mean for AI regulation / Tech Crunch (10 minute read)
AI4Culture — Empowering Cultural Heritage through Artificial Intelligence / ai4culture (4 minute read)
It aims to empower users with the knowledge and skills needed to leverage AI for preserving, managing, and promoting cultural heritage.
tools, datasets, and upskilling materials
AI search could break the web — Developers should act before governments fall back on blunt tools. / MIT Technology Review (11 minute read)
slightly hyperbolic title (aren’t they all?) but interesting nonetheless
📚 FOUNDATIONAL
Understanding Multimodal LLMs / Sebastian Raschka, Substack (sorry) (27 minute read)
What’s the difference between CPUs, GPUs and TPUs? / Google blog (6 minute read)
How I write code using Cursor: A review / Arguing with Algorithms (12 minute read)
both a primer and insight into how A.i. is changing coding
related: GitHub Spark — Can we enable anyone to create or adapt software for themselves, using AI and a fully-managed runtime? / Github Next (10 minute read)
🚀 FOR LEADERS
Exclusive: Walt Disney forms business unit to coordinate use of AI, augmented reality / Reuters (6 minute read)
some companies think it’s not “just IT”
AI in Production Guide / Azure, Github (4 minute read)
Our aim is to demystify the complexities of AI project development in Azure, offering a clear roadmap from conceptualizing AI solutions to their full-scale deployment. By consolidating Azure AI information and tools, we facilitate a smoother and more efficient integration of AI into your business operations.
Artificial Intelligence, Scientific Discovery, and Innovation / Aidan Toner-Rodgers, National Bureau of Economic Research (46 minute read)
AI-assisted researchers discover 44% more materials, resulting in a 39% increase in patent filings and a 17% rise in downstream product innovation. These compounds possess more novel chemical structures and lead to more radical inventions. … Survey evidence reveals that these gains come at a cost, however, as 82% of scientists report reduced satisfaction with their work due to decreased creativity and skill underutilization.
🎓 FOR EDUCATORS
Professor tailored AI tutor to physics course. Engagement doubled. / Harvard Gazette (8 minute read)
$100K or 100 Days: Trade-offs when Pre-Training with Academic Resources / arXiv (48 minute read)
Our results reveal a brighter picture for academic pre-training: for example, although Pythia-1B was originally trained on 64 GPUs for 3 days, we find it is also possible to replicate this model (with the same hyper-parameters) in 3x fewer GPU-days: i.e. on 4 GPUs in 18 days. We conclude with a cost-benefit analysis to help clarify the trade-offs between price and pre-training time. We believe our benchmark will help academic researchers conduct experiments that require training larger models on more data.
📊 FOR TECHNOLOGISTS
json_repair — A python module to repair invalid JSON, commonly used to parse the output of LLMs / mangiucugna, Github
works as advertised
should just be built-in to the LLMs because their JSON frequently has issues, in our experience
Aide — The open-source AI-native IDE.
no one needs to learn another IDE, but worth trying
Making ChatGPT Friends with Numbers — So you can trust its results and work really faster / Tobias Zwingmann (8 minute read)
Strategy #1: Code Generation
Strategy #2: Tool Use
Strategy #3: Rankings
🎉 FOR FUN
Oasis / oasis-model, Github
We’re excited to announce Oasis, the first playable, realtime, open-world AI model. It’s a video game, but entirely generated by AI. Oasis is the first step in our research towards more complex interactive worlds.
demo has 10-minute limit and only runs in Chrome
Holiwise — Personalised Travel Destinations, Itineraries & Hotels
beware of hallucinations, but top three recommendations seemed to be accurate in our hometowns; links to Expedia for accommodations
Prime Video’s X-Ray Recaps Feature Keeps You Up to Speed on Your Favorite Shows / About Amazon blog (6 minute read)
how is this better than skimming wikipedia?
🧿 AI-ADJACENT
How to Build a Platformer with AI - Full Tutorial / Rosebud AI (29 minute read)
⋄