Febuary 27, 2024 🧠 Google's new AI

Thanks for going deeper with us!

Hey folks! It’s Tuesday, and you know what that means. It’s time for the latest edition of Deeper Learning.

In today’s digest:

And don’t forget to tell us what you thought about today’s newsletter at the end!

— The Product Hunt team

P.S. Someone awesome forward you this newsletter? Sign up here.


Google just launched a new lightweight LLM

Google is going the route of open AI. No, not that OpenAI, I mean open-source AI. The tech giant just launched a pair of sister models to its flagship AI Gemini, called Gemma.

Gemma comes in two sizes — Gemma 2B (2 billion parameters) and Gemma 7B (7 billion parameters), which both ship with pre-trained and instruction-tuned variants and are built from the same research that informed Gemini’s development.

According to the announcement, Gemma isn’t meant to replace Gemini, Google’s competitor to GPT-4. Rather, they are designed to be lightweight alternatives suitable for more remedial tasks like chatbots and text summaries. 

Don’t balk at the thought of a more lightweight AI just yet, though. What Gemma lacks in size, it makes up for it in speed and accessibility. The company states that “Gemma 2B and 7B achieve best-in-class performance for their sizes compared to other open models,” and they can even be run directly on your laptop using Hugging Face, Kaggle, and Google’s Vertex AI. 

Why it matters: While developers can build on Gemini through APIs, Gemma is open-sourced, meaning developers have more freedom to experiment with the same research that informed the company’s flagship product and even contribute to its direction. 

Of course, this comes with its own concerns as open models are harder to place guardrails on, so Gemma ships with “responsible AI toolkits,” and the models went under extensive testing by “red-teamers” — experts in things like misinformation, hate content, and bias. 

Developers can use Gemma for free on Kaggle, and first-time Google Cloud users will receive $300 in credits to use the models. In the meantime, I’m placing a bet that the next Google AI will be named Gary. 


Microsoft’s new golden AI

Mistral is a French AI company that launched in May last year. It quickly received unicorn status in December after securing $385 million in funding led by A16z. The company is co-founded by Google DeepMind and Meta alums and is being touted as a potential European rival to OpenAI. 

The team is hoping to make good on that claim by launching a new flagship Large Language Model (LLM) called Mistral Large that is designed to rival some of the top-tier models like GPT-4 and Claude.

Mistral Large is marketed as a cheaper GPT-4 alternative. It’s open-source by design, supports 32k context windows, has native function calling capacities, and is even bilingual from the get-go. 

According to the announcement, Mistral Large achieves fairly strong results on some commonly used benchmarks, to the point that it takes second place ahead of Claude and just behind GPT-4 when you measure multitask language understanding. 

Why does it matter? Well, Microsoft seemingly doesn’t want to put all its eggs in one basket with OpenAI and has announced a multi-year deal with Mistral. The deal will include around $15 million in investment and a research and development program to build applications for governments across Europe on top of Azure. 


Hit inbox zero in no time

Superhuman was founded way back in 2014 to help solve email bankruptcy. Fast forward ten years, and the team is back to launch Superhuman AI 2.0, a suite of features designed to make email even faster. 

Here’s the lowdown: 

Instant Reply crafts three draft replies to each email you receive that you can either edit or hit send and get on with your day. 

Auto Summarize exists to reclaim the time you’d normally spend reading an email. It gives you a quick, one-line summary at the top of every email to quickly gather context and craft a response from the get-go. 

Idea to email is kind of like prompt writing but for email. Similar to, say, the ChatGPT interface, you can now jot down a few phrases, and Superhuman will generate a fully-written email, ready to send. It even matches the tone and voice you’ve used in previous emails. 

Alongside that, Superhuman AI 2.0 also ships with features like an AI editor and one that allows you to write in any language. If you’d like to test it out, the team is offering $60 worth of Superhuman for anyone who signs up today. 


PitchLeague is a personalized AI assistant for writing your pitch deck. 

OpnForm lets you build forms in no time with an AI-powered platform. 

Kraftful 2.0 is an AI toolkit built to help product teams ship faster.

Talently launched an AI assistant that helps you prepare for your next job interview.

buywith is a shopping assistant that lets you engage with brands and assistants. 

Fusion Copilot uses AI to give you detailed insights into your day-to-day activities.

Slite launched a new AI tool that makes your documents much more readable.

Thinkbuddy AI is a native Mac app for running ChatGPT.


Here via forward? Subscribe here.

Join the conversation

or to participate.