Perplexity partnering with publishers

Howdy, wizards. A short one to wrap up the week.

Let’s dive in!

‍

Dario’s Picks

The most important news stories in AI this week

‍

1. Perplexity is partnering with publishers. Following lots of criticism from publishers about Perplexity's aggressive approach to data scraping and minimal attribution of sources, the company is taking a practical step to build “a thriving information ecosystem”. The new Publishers’ Program gives partners revenue-sharing deals, as well as access to their API. The first partners include names like TIME, Fortune, Wordpress, and more.
‎

‎ Why it matters‎ ‎ According to their press release, Perplexity's goal is to «build an internet with aligned financial objectives». That all sounds great - like exactly what we need - but with their approach so far of shooting first, asking for forgiveness later, and only changing things when they get sued, i’m having the feeling we haven’t reached the equilibrium just yet.

‍

2. Anthropic called out for aggressive data scraping. Perplexity isn’t the only company getting accused of breaking the internet's etiquette when it comes to web scraping lately. Some big names like Freelancer.com is claiming Anthropic is sending massive bot armies (3.4 million bots in 4 hours in this case) to scrape their data, and not being able to stop them using standard web protocols. The amount of bots sent by Anthropic is reportedly so high it increases bandwidth charges and slows down the operator’s site significantly.

‍

‎ Why it matters‎ ‎ It’s almost a daily occurrence by now: a new AI company infuriates publishers with their dirty approach. It’s becoming increasingly clear that we’re in a transitional period where AI companies can get away with behavior like this. They’ll likely need to adapt soon, both in terms of how aggressive the scraping is and in terms of revenue sharing.

‍

3. ‎ Google's AI wins the silver medal at the math olympics. Google Deepmind launched 2 AI systems, AlphaProof and AlphaGeometry 2, which together achieved the same level of problem solving as a silver medalist in the Mathematical Olympiad.

‍

‎ ‎ Why it matters‎ ‎ It's not long ago achievements like this was unthinkable, or at least expected further into the future. Google's new AI systems are making big contributions to the field of math, and Google states they're excited about "AI system like Gemini becoming more capable at math and reasoning". This implies that these systems, which aren't GenAI but nevertheless AI, could be somehow integrated with traditional LLMs in the future.

‍