• Ben's Bites
  • Posts
  • Daily Digest: Google's secret stock for AI

Daily Digest: Google's secret stock for AI

PLUS: Open AI attempts at crowdsourcing AI policy.

January 17, 2024

Sign up | Advertise | Ben’s Bites News
Daily Digest #327

Hello folks, here’s what we have today;

PICKS

  1. We're looking at how AI's future is being shaped, and guess what? It's not just tech geeks in the driver's seat anymore. Open AI is creating a new team to get the general public’s input into framing AI policy. OpenAI started with 10 AI grants focusing on different projects around AI policy and governance.🍿Our Summary (also below)

  2. Google is guarding its AI talent from leaving to join OpenAI. Google is tapping a pool of fast-vesting stock grants worth up to millions per person to prevent its key AI talent at DeepMind from leaving for OpenAI.🍿Our Summary (also below)

  3. Bloomberg interviewed Sam Altman (CEO) and Anna Makanju (VP, Global Affairs) from OpenAI at the World Economic Forum in Davos. Among all the hypotheticals and tangents, I found some key statements for you to catch up on.

TOP TOOLS

  • Tiny Narrations by SF Compute - Synthetic audio dataset based on TinyStories.

  • AI assist by Equals - Next-generation Clippy for a next-generation spreadsheet.

  • Book and Bot - A story about Maya’s pet goat. Read the book, then chat with the goat.

  • Pinecone Serverless - Build knowledgeable AI apps at up to 50x lower cost. $100 in free credits.

  • Vercel + Pinecone - You can now use the Pinecone integration to create vector databases for your AI applications in Vercel.

  • Podsqueeze 2.0 - Repurpose your podcast into clips, show notes, blogs and more.

  • Gotalk AI - Easily convert text to natural-sounding AI voices.

  • Ubique - Sales tool to generate personalized videos at scale.

NEWS
QUICK BITES

We're looking at how AI's future is being shaped, and guess what? It's not just tech geeks in the driver's seat anymore. Open AI is creating a new team to get the general public’s input into framing AI policy. OpenAI started with 10 AI grants focusing on different projects around AI policy and governance.

What’s going on here?

OpenAI is mixing things up by using crowdsourced public opinion steer AI. It's pretty cool – your ideas and opinions could help decide how AI works.

What does this mean?

So, these 10 teams from different places and backgrounds are working on some cool stuff. They're not all tech nerds; some know about law, journalism, and other stuff. These projects are pretty varied too.

  • 📚 Case Law for AI Policy: AI interaction case repository

  • 💬 Collective Dialogues: Policies from public deliberation

  • 🤝 Deliberation at Scale: AI-facilitated small group discussions

  • 🦉 Democratic Fine-Tuning: Value elicitation for model tuning

  • ⚡ Energize AI: Guidelines for AI alignment

  • 👫 Generative Social Choice: Representing diverse opinions

  • 🌎 Inclusive.AI: Decentralized decision-making for underserved groups

  • 📰 Making AI Transparent by Rappler: Offline/online discussion linkage

  • 🎨 Ubuntu-AI: Inclusive LLM development for African creatives

  • 🔁 vTaiwan and Chatham House: Recursive participatory processes

To break down the learnings, public opinion on AI is always changing, so we gotta keep tabs on it constantly. Then there's the digital divide – it's tough getting everyone on board when not everyone's online or tech-savvy. There's this tightrope walk between reaching a consensus and respecting diverse viewpoints in decision-making.

Now, OpenAI is putting together a “Collective Alignment” squad. The mission? To weave public input right into the fabric of AI models and give those grant prototypes a real-world test drive. And they are on the lookout for some sharp research engineers to join the team.

Why should I care?

AI's everywhere, right? Your social media, your job, everywhere. But who gets to say what's cool or not for AI? Up until now, it's been mostly tech folks. But now, it's about getting your thoughts in there too. This is important because it's not just about making AI clever. It's about making it something that works for all of us, something that gets our values. We're shaping AI to be a part of our world, growing with us and reflecting what we all care about.

QUICK BITES

Google is using special stock compensation to retain its top AI researchers as OpenAI tries to poach them with huge pay packages.

What is going on here?

Google is guarding its AI talent from leaving to join OpenAI.

What does this mean?

Google is tapping a pool of fast-vesting stock grants worth up to millions per person to prevent its key AI talent at DeepMind from leaving for OpenAI. Google is pulling out all the stops to keep its top AI experts from being lured away by OpenAI's massive compensation offers.

Select DeepMind employees working on Google's Gemini model to rival ChatGPT are being given large grants of Google stock that vest in just 1 year, compared to the normal 4 year vesting period. These grants are in addition to salary and bonuses and can be worth millions based on the person.

Why should I care?

The talent war between Google and OpenAI is crazy and it’s not new. Many key AI researchers at OpenAI have previously worked at Google and led breakthroughs in the AI space. The departures of key personnel could further back Google's efforts to challenge OpenAI, and hence Google is trying to leverage its stock and existing talent pool to stay competitive.

QUICK BITES

Bloomberg interviewed Sam Altman (CEO) and Anna Makanju (VP, Global Affairs) from OpenAI at the World Economic Forum in Davos. Among all the hypotheticals and tangents, I found some key statements for you to catch up. Most of these are slightly paraphrased statements from Sam and a few from Anna.

  • It’s good that we have a lot of anxiety regarding upcoming elections.

  • AGI will be developed soon but it’ll change the world much less than we all think.

  • OpenAI is working with the US military on open-source cybersecurity software and preventing veteran suicide. The recent change in Open AI’s policy that removed the words “military and warfare” was to make these use cases clear. Developing weapons and harmful tools is still prohibited. [additional context here]

  • We generally don’t want publishers’ data for training […] Some people want to partner with us, some don’t. We’d defend the lawsuit but rather have people say no to partnerships than sue us.

  • I am looking at a future where instead of artists deleting/restricting their styles in DallE, they get some economic benefit from being part of that ecosystem. We want the artist to say: “That’s awesome.”

  • For the most part, people are interested in governance and regulation but I’m even more excited about governments using these systems to serve citizens.

  • My view of the world is that there are two important currencies in the future “intelligence” (compute) and “energy”. Meeting the energy requirements for AI will need a breakthrough and that will also be a step in solving climate-related issues.

  • I’ve not heard anything official about a partnership with Johnny (Ive). Speaking about Johnny, he has made computers very human-compatible again and again. OpenAI and the field (AI) haven’t been very good at it.

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

or to participate.