- Big Brain
- Posts
- 🧠 Bloomberg bets big on AI
🧠 Bloomberg bets big on AI
PLUS: Y not AI?
Happy Monday and welcome back to Big Brain!
We hope you had a great weekend and are ready for another action-packed week.
🔦 Today's AI spotlights:
Bloomberg bets big on AI
Y not AI?
Is smaller actually better?
🔦 BLOOMBERG BETS BIG ON AI
What if you could use AI to make yourself rich?
For anyone who's ever thought this, one company is already a step ahead.
Bloomberg has unveiled a new AI model called BloombergGPT. The finance-chatbot is trained on a massive amount of financial data to help with any of the following tasks:
Analyze risk
Judge financial sentiment
Automate accounting and auditing tasks.
Naturally, BloombergGPT will have access to the Bloomberg Terminal. This will give it a huge edge due to the massive amounts of real time data housed in it.
Its engineers also trained the model with general-purpose datasets. Why would a financial model need that?
Well, Bloomberg wants it to also be competitive in performing general tasks.
Just like ChatGPT did for communications, many expect BloombergGPT to revolutionize the finance industry.
🔦 Y NOT AI?
Every year, the startup accelerator Y Combinator (YC) holds two rounds of funding.
Guess how many of the latest round included AI startups?
About 34% of them (91 in total). About 20% of them (54 startups) use generative AI specifically.
This is higher than in previous YC cohorts. Past classes had 44 using AI in general (9 companies of which used generative AI).
Clearly, there's been a change in sentiment.
It should be noted that some startups could be highlighting AI to attract investments in a tough fundraising market.
But don't worry, these startups aren't all talk. Many are using AI in nuanced and exciting ways.
One example is the company Hadrius using it to make sure that documents and business practices are SEC compliant.
Another, CreatorML, predicts how much engagement a YouTube video will get.
No matter how you slice it, one thing is for sure:
AI has the potential to change industries, and YC doesn’t want to miss out on it.
(Fun fact: OpenAI CEO Sam Altman, is the former president of Y-Combinator)
🔦 IS SMALLER ACTUALLY BETTER?
L(arge)LMs, as per their name, require huge amounts of data to operate properly.
What if there were an easier and more efficient way?
Some are beginning to experiment with what they're calling “smaller LLMs” (sLLMs). These models can be tailored to specific companies or, better yet, specific customers.
Here's the idea:
Each company has its own lexicon with customer data and content. It can use it to train its own sLLM to generate content that is more customized.
Still, building sLLMs can pose many challenges, especially when it comes to users' private data.
This hasn't stopped Databricks, a "cloud data lakehouse" startup, from having a go at it.
It recently released Dolly, an sLLM that's allegedly more accurate and focused than LLMs like ChatGPT.
Since it's trained on less data than the latter, Dolly and sLLMs like it are also much cheaper to produce.
While promising, Cisco's Jeetu Patel thinks that sLLMs may not be the future.
This is because companies also have the option of feeding its users' data into LLMs to garner the same result.
📰 OTHER HEADLINES & STORIES
🛠️ AWESOME AI TOOLS
CreatorGPT: 1000+ GPT4 prompts with 10+ use cases for content creators
MyGPT by SamurAI: Access to plugins without ChatGPT Plus
Cheat Layer: Automate your business from end to end in natural language
UnPrompt: Get stable diffusion prompts for any image, subject or style
🖼️ AI ART PIECE
Yellowstone Gucci — by Nick St. Pierre
🤖 CHATGPT PROMPT OF THE DAY
Rumor has it, this was why ChatGPT was banned in Italy 😂 (via Reddit)
How would you rate today's newsletter? |