- The Signal
- Posts
- Sunday Signal: Databricks’ new LLM, the first 15 seconds and stolen focus
Sunday Signal: Databricks’ new LLM, the first 15 seconds and stolen focus
Alex Banks
March 31, 2024
Hey friends 👋 Happy Sunday.
Here’s your weekly dose of AI and introspection.
Today’s Sunday Signal is brought to you by Drippi AI.
Drippi is an AI assistant for your Twitter DMs. It helps you scrape thousands of high-quality leads in seconds, automate personalised outreach messages in minutes and tame your inbox with filters, auto-replies, and more.
Want to feature here? Sponsor this newsletter.
AI Highlights
Databricks unveils DBRX, an open LLM, outperforming established models like GPT-3.5 and comparable with Gemini 1.0 Pro. DBRX shines in programming, surpassing CodeLLaMA-70B, with significant efficiency gains: it's 40% smaller than Grok-1 and boasts 2x faster inference than LLaMA2-70B.
Alex’s take: I was impressed with DBRX’s compute efficiency, which is nearly 4x better than previous models. What’s more, you can try it out on Hugging Face and You.com.
In September of 2023, Amazon announced they would invest up to $4 billion in Anthropic, the rival to OpenAI’s ChatGPT. Wednesday’s news marks Amazon’s second tranche of that funding.
Alex’s take: The cheque signified Amazon’s largest outside investment in its three-decade history. It makes me ask again the question I posed a few weeks ago: who will build the best chatbot by the end of 2024? With Claude recently overtaking GPT-4 in the chatbot arena leaderboard, the competition is indeed heating up—and Amazon seems to be in a good place to capitalise on the upside.
Hume introduces the first conversational AI with emotional intelligence. EVI understands the user’s tone of voice, adding meaning to every word and uses it to guide its own language and speech.
Alex’s take: Intimate, always-on AI assistants were once a Hollywood fantasy. However, current speech-to-text technology has been stuck, feeling inauthentic and forced. With the arrival of emotionally aware AI, you can now talk to it like you would to a human, opening the door to some really exciting (and potentially dangerous) use cases.
1 Article I Enjoyed
Scott Belsky is an American entrepreneur, author, and early-stage investor best known for co-creating the online portfolio platform Behance in 2006. He served as CEO until Adobe acquired Behance in 2012.
His 2013 article helps answer the burning question, “How do you build a product that engages a user quickly enough to engage them over time?”
My favourite takeaways:
In the first 15 seconds of every new experience, prospective customers are all lazy, vain, and selfish.
A ‘hook’ pulls us past our laziness, vanity, and selfishness.
The science of cover design is entirely different from the science of great writing.
For customers you’ve hooked, reeled in and survived the first 15 seconds, build a meaningful experience and relationship that lasts a lifetime. You must bifurcate your approach.
1 Idea I Learned
We live in an attentional pathogenic culture.
This idea is from Johann Hari’s book, “Stolen Focus”.
It describes an environment in which sustained and deep focus is extremely difficult for everyone, and you have to swim upstream to achieve it.
It is not your fault you cannot focus.
Most office workers never get an hour to themselves without being interrupted. For example, the average CEO of a Fortune 500 company gets just twenty-eight uninterrupted minutes a day.
Three ways task switching degrades your focus:
Switching cost — You lose focus switching between tasks.
Screw-up effect — When you switch between tasks, errors that wouldn’t have happened otherwise start to creep in.
Creativity drain — You will likely be much less creative in the long run.
We must protect our focus, for solving big problems requires the sustained focus of many people over many years.
JFK on doing hard things:
“We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard.”
I find Patrick Collison’s interpretation of the quote to be hugely applicable to startup founders.
“We do these things not because they are easy, but because we thought they were going to be easy.”
Only people who are naturally somewhat overconfident will take on problems that, when viewed rationally, would seem too hard.
As Elon Musk stated, “I thought SpaceX and Tesla both had >90% chance of failure, but worth trying anyway.”
You have to be slightly naive to do bold things. Especially as a founder when you’re solving hard problems and the odds are stacked against you. A rational mind would see the tremendous task ahead and dismiss it. Therefore, doing must precede understanding.
1 Question to Ponder
Can you imagine yourself in 10 years if, instead of avoiding the things you know you should do, you actually do them every single day?
💡 If you enjoyed this issue, share it with a friend.
See you next week,
Alex Banks