The Neuron
Posts
😺 McKinsey's legendary chatbot

😺 McKinsey's legendary chatbot

PLUS: We're (not) like The New York Times

Noah Edelman & Pete Huang
August 17, 2023

Welcome, humans.

Keep your eyes peeled for a tasty rebrand coming soon. Spoiler alert: our “riding solo” orange tabby is about to meet up with an entire village of felines! 😺😸😹😻😼😽🙀😿😾

Here’s what’s going on in AI today:

McKinsey built a chatbot trained on all its knowledge.
AP News & The NYT are approaching AI with caution.
Google is launching its OpenAI competitor “Gemini” this fall.
Google is building an AI for “personal advice”.

Sponsor The Neuron | New Here? Subscribe!

McKinsey’s New AI Can Answer Any Consulting Question.

McKinsey built an internal chatbot called “Lilli” that’s trained on the company’s entire knowledge base (literally >100k pages of documents).

Lilli can give employees expert advice on anything project-related: research, data, relevant cases, & more.

In other words, every McKinsey employee now has the capacity to "think" like the most brilliant McKinsey consultant ever.

Reminds us of an April study where average customer support agents learned to perform like senior customer support agents using AI.

Why it matters: An AI that knows everything about your business will be a HUGE productivity unlock. Lilli already helps McKinsey staff cut down time spent on research from weeks → hours.

But to build a chatbot like Lilli, your business's knowledge must be retrievable. This means having it all written down. In one place. Most businesses’ knowledge is stored all over the place … in Notion, on notes, and in their noggins.

We recommend companies dedicate the next few months cleaning up all their resources so it they can train an AI on it.

FROM OUR PARTNERS

Use AI To Trade Like A Wall Street Titan.

Wall Street legend Jim Simons has generated 66% returns annually for over 30 years. His secret?

Algorithmic trading (and also being a genius at math and coding).

With Composer, you can create algorithmic trading strategies that automatically trade for you (no coding required).

Build the strategy using AI, a no-code editor, and 1000+ community-built strategies.
Backtest against other stocks & ETFs using professional grade data.
Start trading automatically with the click of a button.

Composer’s got big shots on board, with over $1 billion in trading volume (and the slickest landing page we’ve ever seen).

Join the Composer community and unlock the power of algorithmic trading!

past performance is not indicative of future returns.

Traditional Media Outlets Are Scared Of AI.

Household media outlets are straight-up in a state of panic about how to handle AI. Vanity Fair reported that The New York Times spent the entire summer “freaking out” over AI.

Now, they want to take control of AI before it gains the upper hand. The Associated Press (the standard setter in journalism practices) is out with new guidance re: AI:

Verify facts from any AI
Don’t put sensitive info into a chatbot
Make sure your sources don’t use AI content.

Here’s our take: We agree that anyone publishing content should be careful about what an AI tells them. But dismissing AI completely in the process feels like the 2023 version of stubbornly clinging to a typewriter (as cool as they are).

We’re personally using AI to generate ideas and wordsmith sentences. And yes: we’re borderline-OCD about ensuring the facts are facts before clicking publish.

Around the Horn 🦄

Google will launch “Gemini,” its OpenAI competitor, this fall.
Snap’s “My AI” glitched yesterday, posting on its own Story.
Replit added enhancements to its AI pair programmer “Ghostwriter.”
Google is developing an AI model capable of providing 21 types of personal and professional advice.
OpenAI acquired an AI design studio named Global Illumination.