Weekly Link Roundup
Interesting Links or Signals from the Future? You decide.
New Lenovo ThinkPad X1 Carbon hints at new 2024 trend: AI laptops: How’s that enterprise-wide laptop refresh rate and budget looking?
The Year That Twitter Died: This is a great read about what used to be a great site. I think my largest follower count on Twitter was around 7K but wow did I pump out a lot of tweets there. It was my cognitive attic.
Julian Stodd, Geoff Stead, and Sae Schatz have published a great inquiry into questions around #GenAI. I somehow snuck in as a tiny contributor with other folks like Donald Clark and Mark Zao-Sanders. Links to the book are here including a FREE PDF version.
Midjourney Alpha is here with AI image generations on the web: Makes me a little sad. I like having my little MJ server in Discord.
Visual Electric launches to liberate AI art generation from chat interfaces: Here is the absolute key “This is because Dunn and his two co-founders — Visual Electric chief product officer Adam Menges, formerly co-founder of Microsoft acquisition Lobe; and chief technology officer Zach Stiggelbout, also formerly of Lobe — believe that image generation AI models are in the process of being commoditized, and that it is the front-end user interface that will most differentiate companies and separate the successes from the failures.”>The differentiation will be front end.
Why It’s So Hard to Search Your Email: I hate searching my email but I do love stories that break down what we see as monolithic “things” like search > “It is genuinely harder to sort and rank 10 gigabytes of your emails than it is to sort all of the websites in the world.”
OpenAI’s nonprofit arm showed revenue of $45,000 last year, even though company is worth billions: “OpenAI’s nonprofit organization generated $44,485 in revenue in 2022, almost entirely from investment income, according to a document filed with the Internal Revenue Service. The filing doesn’t disclose revenue from OpenAI’s for-profit division, which is commercializing ChatGPT.”
Microsoft releases Phi-2, a small language model AI that outperforms Llama 2, Mistral 7B: Small enough to run on a laptop but with “its 2.7 billion parameters (connections between artificial neurons) boasts performance that is comparable to other, much larger models including Meta’s Llama 2-7B with its 7 billion parameters and even Mistral-7B, another 7 billion parameter model.”
Beyond Buzzwords: Glasswing AI palette guides startups navigating AI’s diverse terrain: Couple of good lessons in here. 1st lesson - in a gold rush, its the people selling picks and shovels who get wealthy. 2nd lesson “The problem with LLM wrappers? They’re indefensible.” Yup.
Teens, Social Media and Technology 2023 (Pew Research Center): “YouTube, TikTok, Snapchat and Instagram remain the most widely used online platforms among U.S. teens.” >These stories/surveys are important even if teens aren’t your customers because they will be soon and oh yeah, they’ll be your employees soon too and they’ll bring these expectations with them.
The Met Will Return 16 Ancient Treasures Tied to Looting: Good
DeepMind’s AI finds new solution to decades-old math puzzle — outsmarting humans: I really dislike the way headlines are written. An LLM (by definition pre-trained) discovered a novel solution. No humans were outsmarted. In fact, without decades of work BY HUMANS on the problem, there would’ve been no data set to train on. I do think this is a positive step > “What also makes the tool quite promising for scientists is that it outputs programs that reveal how its solutions are constructed, rather than just what the solutions are.”
‘Green Roads’ Are Plowing Ahead, Buffering Drought and Floods: Potentially huge - imagine re-engineering roads to help vs hurt the environment. “In the Ethiopian state of Tigray, green road techniques raised the water table and improved the productivity of adjacent farms by 35 percent.”
The first Core Ultra handheld gaming PC is already coming: I get that some folks reading this don’t play computer/video games BUT I share stories like this since games are typically the most demanding programs and if any machine can run the latest games, they can probably run anything.
In a World First, a Patient’s Antibody Cells Were Just Genetically Engineered: “Our B cells help prevent us from getting sick. Their job is to make antibodies, immune system proteins that fight off viruses and other foreign invaders. And they make a lot of antibodies—thousands of them every second. What if these antibody factories could be harnessed to make other things the body needs? That’s the idea behind a trial launched by Seattle-based biotech company Immusoft. The company announced today that its scientists have genetically programmed a patient’s B cells and put them back in his body in an effort to treat disease. It’s the first time engineered B cells have been tested in a person.”
The Stanford Prison Experiment was massively influential. We just learned it was a fraud: Please tell me that we all knew this already.
OpenAI’s Ilya Sutskever Has a Plan for Keeping Super-Intelligent AI in Check: Can I just say how HAPPY I am that this team, built to discover ways to keep AI from doing bad stuff, is being stood up A YEAR AFTER the release of ChatGPT. Right on Silicon Valley time. >Lots of places struggling with this, ex post facto.
Cheating Fears Over Chatbots Were Overblown, New Research Suggests: Say it again, we overestimate in the short run and underestimate in the long run.