Bot Eat Brain
Posts
Sam Altman testifies in Congress

Sam Altman testifies in Congress

PLUS: AI tries 20 jobs

Hilbert Publications & Anthony Castrio
May 17, 2023

Good morning human brains, welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

Open AI wants regulation 🎙️
CEO Sam Altman’s AI safety hearing at the US Senate.
AI in the military 📈
ScaleAI announced AI products for US defense.
AI tries out 20 jobs 🧑‍🏭
Wired experimented to see what jobs ChatGPT can replace.

Sponsor Bot Eat Brain | New here? Subscribe!

APPETIZER

Sam Altman Testifies 🏫

Yesterday afternoon, Sam Altman, CEO of OpenAI, testified before the US Senate panel alongside an acclaimed NYU prof and IBM’s AI governance rep.

Here are the key takeaways:

OpenAI CEO testifying before Congress.

1/ Both IBM and Altman urged lawmakers to regulate AI, calling the current boom a "printing press moment" that requires safeguards at the federal level.

2/ Sam Altman confirmed that he’s got zero equity staked in OpenAI. The upside from the seven-year gruel and OpenAI’s $30B valuation? He shares none of it. In his words, “I’m doing it because I love it.”

3/ Sam emphasized the use of AI to manipulate voters and mass attacks with disinformation as his most serious short-term concerns.

"I can’t remember when we’ve had companies come before us and plead with us to regulate them. ‘Stop me before I innovate again.’”

US Senator Peter Welch

4/ Altman suggested the US government regulate the industry by creating a licensing regime for the most powerful AI systems. If it goes through, it’ll mean you need government approval & safety checks to build your own LLMs.

5/ Sam addressed the impact of AI on jobs, saying it’ll require a partnership between industry and government to mitigate the damage, but OpenAI predicts we’ll find new, better ways to be productive at the end of the road.

OpenAI’s public call for regulation comes right after big tech’s White House hearing and Elon Musk’s open letter to pause AI development that’s slowly fizzled out of relevance.

The bigger picture: ChatGPT’s virality + the tsunami of tech companies pivoting to focus on AI + Europe drafting its own legislation has got US lawmakers serious. This is the first in many meetings to draft new guardrails for AI development.

AI regulation & licensing - yay or nay?

BUZZWORD OF THE DAY

Fine Tuning

Additional training provided to an AI model that makes it better at a particular task.

AI models like ChatGPT are built for general purpose functions. They’re pretty good at many things, but rarely experts at any one thing.

Fine-tuning allows you to take a general-purpose model and make it much better at a particular task you care a lot about.

Example: a version of ChatGPT fine-tuned on medical data would become much better at giving medical advice.

WITH PIXIEBRIX

Bend any website to your will with this AI-powered Chrome extension.

PixieBrix lets you customize, extend, and integrate your favorite SaaS apps without writing a single line of code.

Choose from pre-made mods that integrate with hundreds of applications and platforms or build your own to customize your personal experience.

Why it’s cool:

It’s a Chrome extension, so it’s dead simple to install.
Low-code "bricks" get you started fast 🧱
Context-aware, PixieBricks knows what’s up on any page you visit.
Works seamlessly between internal and external apps.

MAIN COURSE

ScaleAI’s Military AI 🪖

Five years ago, 19-year-old Alexandr Wang founded ScaleAI to label big data for AI development. Two years ago he became the youngest self-made billionaire. Last week, ScaleAI revealed an AI system built for the US military.

Donovan is an AI-powered decision-making hub designed specifically with governments in mind and the US Army's 18th Airborne Corps has already deployed Donovan for real-time AI-powered decision-making.

The Donovan Demo

Why this matters: both intelligence and defense agencies collect massive amounts of data but struggle to build systems that can make sense of it all.

If Scale can help government agencies clean, segment, and organize their data, it’s going to play a huge role in national security over the next few decades (and put them in direct competition with companies like Palantir.

One more thing: ScaleAI just launched their Enterprise AI Platform to help big industries bring AI on-premise, protect internal data, and fine-tune models for their needs. Fox Corp and Koch Industries are already customers.

ScaleAI’s ambitions are clear: become the next AI behemoth.