December 12, 2023: Meta's groundbreaking demo

Author

Product Hunt
December 12, 2023

Thanks for going deeper with us!

This was a (mostly) great week for someone other than OpenAI and Microsoft.

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.

SECOND FIDDLE GETS A SOLO

Google finally unveils Gemini

Ever since AI hit the mainstream late last year, it’s safe to say that Google has been playing second fiddle to OpenAI and Microsoft. Until yesterday, it seems. Google pulled the curtains back on its highly anticipated AI model, Gemini.

Gemini is Google’s latest large language model (LLM), and it’s been in the works for quite some time. Its release date was somewhat of a mystery, and there were even reports last week that it was delayed until 2024. It’s been described by CEO Sundar Pichai as an AI that will “revolutionize generative technology for business and daily life.”

Gemini comes in three different flavors:

  • Gemini Nano: This is the baby. Built primarily for mobile devices. As of yesterday, Google Pixel 8 Pro users can enlist it to help with things like audio summaries.

  • Gemini Pro: This is the company’s midsize offering and is designed for more complex tasks akin to ChatGPT. It’s currently powering Bard, Google’s alternative to ChatGPT.

  • Gemini Ultra: Finally, this is the powerhouse version, designed to be used with things like data centers and large enterprises. It will launch in 2024 and will power “Bard Advanced,” an upcoming chatbot that will be able to process text, video, images, and more.

AI pundits here and there have been quick to resign Gemini to the back seat, but Bard users are excited about Gemini and have demoed its skills. Check one of them out below.

OOPS

ChatGPT: Will work for tips

Lazy ol’ ChatGPT. If you’ve noticed that ChatGPT isn’t as helpful as it was before, it’s not in your head. Users have reported that the AI bot has been refusing to answer questions or has been “lazy” with the response. Here’s an example that got a lot of attention on Reddit.

Then OpenAI confirmed the issue and said it was looking into it. Meanwhile, a theory that ChatGPT had learned to do less work while “on winter break” has gained steam.

Tip your AI bots. Or offer to at least? 🙃 A ChatGPT user ran an experiment where he gave the bot a prompt and offered to tip for the response. The prompts with tips were longer and more detailed, even though ChatGPT would not actually accept payment.

So, if you want better results from ChatGPT, be prepared to pretend pay up. To be fair, tipping is a good source of motivation when you’re feeling lazy.

Cone of shame for Google. We’re not here to pick on OpenAI though. Google’s exciting start to the week with the launch of Gemini has been somewhat tarnished by controversy around the demo video, which was edited but made to feel like the bot was responding in real-time. Now Google is the butt of quite a few jokes online.

WOAH!

Meta’s new model makes audio from scratch

https://ai.meta.com/blog/audiobox-generating-audio-voice-natural-language-prompts/

On Monday, Meta dropped its successor to Voicebox called Audiobox. The new generative AI model lets people use natural language to describe a sound or type of speech they want to generate.

For example, give Audiobox the prompt “A running river and birds chirping” and that’s what you’ll get. You can even generate a voice from scratch.

The launch marks a groundbreaking leap in generative AI for audio, with Meta saying:

“To our knowledge, Audiobox is the first model to enable dual input (voice prompts and text description prompts) for freeform voice restyling."

MORE NEW TOOLS

PRODUCTIVITY

ENGINEERING, PRODUCT, DESIGN

MARKETING

AI FOR FUN

KIDS

UNTIL NEXT TIME

Written by Sarah Wright with Aaron O’Leary.

Here via forward? Subscribe here.

Did you enjoy today's newsletter?

Login or Subscribe to participate in polls.

Reply

or to participate.