- Ben's Bites
- Posts
- Daily Digest: Vintage data, brand new money.
Daily Digest: Vintage data, brand new money.
PLUS: AI that understands DNA
Subscribe | Ben’s Bites Pro | Ben’s Bites News
Daily Digest #357
Want to get in front of 100k AI enthusiasts? Work with us here
Hello folks, here’s what we have today;
PICKS
Tumblr getting in bed with AI bigwigs. Automattic, the company behind Tumblr and WordPress, is reportedly close to deals with OpenAI and Midjourney. Don’t @ me but internal reports they have already scraped an initial dataset of Tumblr posts from 2014-2023, with stuff that’s not so public.🍿Our Summary (also below)
Google's got a shiny new AI toy for publishers, and they're paying news outlets to play with it. Why? To get feedback and insights on how these tools can streamline workflows. Google says it is to help publishers and journalists with less cash to splash.🍿Our Summary (also below)
Together AI and Arc Institute create Evo - A foundational model for biology that can create small molecules to long genomes. Similar to long context in language models, the ability to understand small changes (single nucleotide) in that long context (genome-scale) makes it useful. The model weights and dataset are open source.
from our sponsor
Meet Myko - Your Data Co-pilot
Tired of building Salesforce reports or waiting for your team to send you the right data?
Myko is the conversational AI for sales and revenue data.
Trained on your specific KPIs, Myko can help your teams focus on driving revenue instead of struggling with reporting.
Get started today for free at myko.ai
TOP TOOLS
Lip Sync is now in early access for Pika Labs’ Pro users.
Playground v2.5 by Playground AI - New foundation model to create images with open weights.
Pi by Inflection AI is now available as a desktop app (both Windows and Mac).
AI graph editing in Julius - Tweak or customize your graphs with natural language.
Instant Reply in Superhuman - Every email has a draft reply.
HostAI - Automate boring vacation rental duties and focus on 5-star hospitality.
Tusk - AI-created pull requests for annoying tickets.
TypePrompt - AI hooks that make your content impossible to ignore.
NEWS
Klarna AI assistant handles two-thirds of customer service chats in its first month. That’s 2.3M chats which would’ve needed 700 full-time agents.
Microsoft announced its AI access principles at MWC.
Google CEO calls AI tool’s controversial responses ‘completely unacceptable’.
OpenAI’s court filing claims that NYT “hacked” OpenAI’s systems. (more like “misused”)
A new paper from Microsoft optimizes LLMs in 1.58 bits.
Money’s flowing:
Glean raises $200M at a $2.2B valuation. I did a deep dive on 🍿how Glean built a $Bn business.
Parspec raises $11.5M seed for bringing AI to construction.
Photoroom is now worth $500M with its $43M series B. (we covered that it’s raising in January)
Inkitt raises $37M series C from Khosla Ventures.
QUICK BITES
Tumblr getting in bed with AI bigwigs. Automattic, the company behind Tumblr and WordPress, is reportedly close to deals with OpenAI and Midjourney.
What is going on here?
Tumblr and WordPress are joining the AI auction with their massive data pile.
What does this mean?
The deal has been buzzing within the Tumblr community for a while now. 404 Media reported that Automattic plans to introduce an opt-out setting for users who don't want their content used by AI companies.
Automattic took the report personally and dropped a statement on "Protecting User Choice". First, it makes a pinky promise about blocking AI crawlers by default and only sharing public stuff where users are chill with it. Then we get to the good stuff: the company is working with unnamed AI companies that respect what the community wants—attribution, opt-outs, and control.
Don’t @ me but that 404 Media report says they have already scraped an initial dataset of Tumblr posts from 2014-2023, with stuff that’s not so public.
Why should I care?
My post about Reddit’s $60M deal for its data aged like a fine wine. More content platforms are realizing they’ve got the goods big AI companies want. Many don’t want to train their own LLMs but licensing this data can mean a big payday. Especially when many are in a cash crunch.
On the other hand, AI trained on this data is gonna be wild. This will be us millennials’ retribution.
QUICK BITES
Google's got a shiny new AI toy for publishers, and they're paying news outlets to play with it. Why? To get feedback and insights on how these tools can streamline workflows. Google says it is to help publishers and journalists with less cash to splash.
What is going on here?
Google's testing out AI that helps publishers churn out content fast (like, scary fast). In exchange for feedback, they're bankrolling news outlets to use the tool.
What does this mean?
Here's how it works: The AI gobbles up reports and news articles from all over the place (government websites, other outlets, you name it). Then, it mashes them all up and spits out snappy summaries in news story format.
To keep things accurate, it uses a cheeky colour-coded system to show which bits mirror the original (yellow is spot-on, then blue, then red for the least). Of course, a human editor still needs to fact-check and give it the once-over.
The deal is for news outlets to publish three stories every day with this tool and one marketing campaign every month for a five-figure payday.
Why should I care?
The first thought is that original sources might lose traffic but there’s always another side of the coin. Imagine smaller outlets pumping out news like the big guys—creating a more level ground in news reporting.
Google claims it's not about replacing journalists, just streamlining but we'll see about that. But, there's bound to be debate about the ethics of it all. Scraping content without permission is shady. Where does "AI-assisted" end and "AI-generated" begin?
Ben’s Bites Insights
We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;
All 10k+ links we’ve covered, easily filterable (1 referral)
6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)
Reply