Hello and welcome from sunny Austin. I spent the last day or so exploring the city and bumped into someone I went to university with, in a packed bar. Bizarre!

This is Ben's Bites your personal AI news bulletin. Today's a long one, don't stress - it's ok to miss a few links. We'll be back tomorrow 😊 

Let's get to it.

Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 20k subscribers? Check out our sponsor options¹⁴⁶.

Join Discord⁷³

🤌 Ben's Picks

1/

I built a vintage image colorizer & restorer using Zapier & Replicate AI models⁷²³

I spoke with Michael last week and tasked him with coming up with an AI workflow idea to build without code and this is what he came out with. There's a step-by-step video and text tutorial and it's awesome. Learning this kind of process is adaptable to almost any AI model (whisper, stable diffusion, etc) so the possibilities are endless. Give it a go and build something - then tag me (@bentossell) so I can show it off here 😊 

2/

OpenAI hiring thousands to write code for generative coding tool⁹⁸⁴

How many unpaid coding interviews does it take to build up a training set for a coding bot? That’s not the start of a joke - OpenAI are on a global hiring spree⁹⁸⁴, as reported by Semafor, to get hold of low-cost labour to conduct basic coding tasks, with unpaid coding tests lasting up to 5 hours. The end goal? Most likely to build up a robust training set of descriptive code/debugging to develop truly powerful coding co-pilots. Now I understand the tweet that Andrej Karpathy sent out last week: ”The hottest new programming language is English”.

3/

Multi·ON: Web browsing co-pilot using ChatGPT^1,342

Does anyone else find it kinda unnerving when you give control of your computer to a tech support person and their mouse is jumping around your page working its magic? Well you’ll find it even spookier with the Multi·ON plugin^1,342, which is exactly that except it’s ChatGPT doing your tasks on your actual browser. In this example⁴⁷¹ ChatGPT successfully gets hold of the world’s first AI-ordered burger using Doordash in Palo Alto - as if I needed any more temptations to add to my Christmas pounds this early in 2023!

It’s not clear from the demos how complicated the tasks can get, with the examples being pretty basic things like navigating well-known websites, posting tweets, and following Sam Altman. So it’ll be cool to see how this develops with time, such as being able to execute the statements without having to watch it click around.

4/

Generative AI finds proteins that kill bacteria⁵⁰⁵

The transformational, game-changing advances in biology made over the past year using AI technology have successfully been pulled into 2023. New Scientist reports⁵⁰⁵ that researchers using generative AI have been able to identify new antimicrobial protein structures that were able to kill E. coli bacteria, opening up huge potential for new drug development. They trained large language models to learn the grammar of how amino acids are combined to form proteins - wow!

Rewardful is an affiliate software for Stripe and Paddle (Sponsor)Just connect your payment provider and let it track referrals, discounts, and commissions for you. Trusted by AI-driven SaaS companies, including Copy.ai, Rytr.me, Synthesia.io, and many more!Check it out here.¹⁰⁴

🛠️ Cool Tools

• Read Pilot - Analyses online articles and generates Q&A cards for you for intelligent reading. (link⁶⁷⁹)

• Make Log - Transform product updates into your most powerful asset. (link⁴²⁷)

• MagicSlides - Creating AI powered presentations in seconds. (link⁸⁴³)

• Ambient - Use AI to instantly convert what's on your screen into a calendar invite. (link³⁶⁷)

• Kadoa - Effortless data extraction in seconds. (link⁶⁵⁵)

• HoustonAI - Experimental AI assistant trained on the Astro docs to answer questions about Astro 2.0. (link²³¹)

• Cova - Supercharge your job search with AI generated cover letters. (link³²⁸)

• MultiOn - Your AI web co-pilot powered by ChatGPT. (link^1,342)

• Mirage introduces AI powered 3D search on Objaverse, the largest annotated 3D dataset. (link²⁴⁴)

• Diff models from Carper AI - Generate changes to code described by a commit message. (link¹⁶³)

• ChatGPT Plagiarism - Find out if your students are using ChatGPT or other AI for their assignments. (link^354†)

• Langchain Chat Next JS - Open source frontend for a versatile Next.js chat interface on Langchain documentation. (link²⁰⁸)

• Email Helper - Generate your business emails in seconds. (link⁴⁰⁸)

• Fuzzy.js - An experimental approach to using GPT-3 as a back-end, communicated via HTML attributes. (link²²⁹)

• Glow AI - Generate your next skincare routine in seconds. (link⁴⁵³)

• Camira - Unlock the full potential of your photography and videography with CAMIRA's AI-powered tools. (link³⁷³)

• Article Summary - Create a short summary of any article to quickly share the main points. (link⁴⁶⁴)

• Shoppy AI - A bot taught using Shopify's help centre documentation + embeddings. (link²⁶²)

• ColBERT adds index updater and base models feature to their state of the art neural search. (link¹⁷⁷)

• Tailwind CSS Answer Bot - Ask anything about Tailwind CSS documentation. (link²⁰⁸)

• Detangle - Summarise any media—text, audio, or video. (link³²⁸) They have a waitlist for new users but don’t worry, we gotchu. Special offer for Ben’s Bites readers to get bumped up the list using this link⁶⁹² today.

🤓 Miscellaneous

• OpenAI has hired an army of contractors to make basic coding obsolete. (link⁹⁸⁴)

• The state of generative AI in 7 charts. (link²⁰¹, without paywall here⁶²⁴)

• Quick thoughts on ChatGPT and why it matters for SaaS startups. (link²⁹⁸)

• AI has designed bacteria-killing proteins from scratch – and they work. (link⁵⁰⁵)

• Natural language is the lazy user interface. (link²¹¹)

• Hugging Face is introducing webhooks. (link¹⁸⁷) Open for beta-testing on request here⁹⁰.

• Two AI startups without revenue look to raise at unicorn valuations. (link³³⁹)

• Permission to be excited about AI. (link⁷⁸, without paywall here²²⁷)

• OpenAI, Microsoft want court to toss lawsuit accusing them of abusing open-source code. (link⁹⁴)

• Big Tech was moving cautiously on AI. Then came ChatGPT. (link^82†, without paywall here¹⁶¹)

• Can generative AI’s stimulating powers extend to productivity? (link⁷⁵, without paywall here¹⁷⁸)

• Artificial intelligence pioneers back $550mn fund for AI startups. (link⁷⁵, without paywall here¹⁸¹)

• Kiss/Crash: A multi-screen installation using AI imagery to explore the theme of desire. (link²³³)

• A prosthesis for imagination - Using AI to boost your creativity. (link²⁵⁰)

• The AI moment, AI vs crypto, & a heated debate with Amjad Masad, Flo Crivello, & Nathan Labenz. (link¹⁸⁵)

• 96th JPEG meeting - Analysis of the responses of two calls for proposals for machine learning-based codec solutions. (link⁶⁷)

• Cool things that happened at Scale AI’s hackathon. (link²²⁶)

• The peak of AI inflated expectations. (link²⁰¹)

🎓 Learn

• Build your own AI based text translator chat. (link¹⁸¹)

• Querying NBA stats with GPT-3 + Statmuse + Langchain. (link¹⁹⁶)

• Effortless Essays - AI-powered workflow to turn voice memos into 1st draft essays. (link¹⁹⁹)

👋 Too many links?! I created a database for all links mentioned in these emails. Refer 1 friend using this link and I'll send over the link database.

🔬 Research

• Noise2Music - A series of diffusion models trained to generate high-quality 30-second music clips from text prompts. (link¹⁵⁸)

• Make-An-Audio: Text-to-audio generation with prompt-enhanced diffusion models. (link¹³⁶)

• Official implementation of Tune-A-Video - One-shot tuning of image diffusion models for text-to-video generation. (link⁹⁹)

• 3DHumans - High-fidelity 3D scans of people in diverse clothing styles. (link¹⁶⁷)

• ChatGPT goes to law school. (link¹³⁷)

• 3D shape representation for neural fields designed for generative diffusion models. (link⁷³)

• PLay - Generating parametrically conditioned layouts in vector graphic space with latent diffusion. (link⁹⁵)

• Accelerating guided sampling process in diffusion models based on operator splitting methods. (link⁵⁵)

• Moûsai: Text-to-music generation with long-context latent diffusion. (link¹¹⁷)

• Leveraging the third dimension in contrastive learning. (link⁶⁷)

• Pre-training for speech translation - CTC meets optimal transport. (link⁷⁶)

• Call for papers - The BabyLM Challenge for sample-efficient pre-training. (link⁶⁶)

📰 Unclassifieds

• AutomationTown - A scripted, bingeable pod about automation & AI, boasting a cast of professional actors. Listen to the season 3 trailer>>⁷⁶

Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 20k subscribers? Check out our sponsor options¹⁴⁶.

🖼 AI images of the day

Everyone listening to AI generated music next month