Gm friends, the weekend is here. Happy getting shit-faced. Going catch you’ll upto the cool shit that happened this week

 _{Big Tech AI Roundup}

There were so many impressive and important AI product launches and news stories from Big Tech this week, that we’re consolidating them all into one roundup. Hold on tight.

OpenAI

There were three major launches or stories from OpenAI this week.

1. ChatGPT can now see, hear, and speak. You can now interact with ChatGPT in a number of ways. The new voice and image capabilities give the product a more intuitive interface and expand its use cases.

   

2. ChatGPT can now search the web in real time. OpenAI resolved one glaring product deficiency this week: ChatGPT can now browse the internet to provide you with current and authoritative information, complete with direct links to sources. It is no longer limited to data before September 2021. Mazee

3. OpenAI, Jony Ive in talks to raise $1 billion from SoftBank for AI device venture. The company is also in talks with famed Apple designer, Jony Ive, and SoftBank to commercial "iPhone of artificial intelligence." Now if that’s not a headline that doesn’t catch your eye, I’m not sure what will.

look at Sam Altman taking over the world right now lol

Meta

Not to be outdone, Meta had a big day of releases and updates at Meta Connect 2023. TechCrunch has the full roundup. Something tells me we’re not going to be using metaverse as pejorative in a couple of years.

1. Quest3: Meta’s new high-res VR headset.

2. Cloud Gaming: The big announcement here was a virtual screen that can float in either a virtual or mixed-reality space

3. Emu: Cute little generative AI stickers that will be available in Meta’s messaging apps. Translate prompts into unique emojis

4. Ray-Ban Smart Glasses: Livestream from your glasses, interact with Meta AI voice assistant, plus all the stuff you could do on earlier versions. Text translation coming soon…

5. Meta AI: Al was the word of the day, across all of Meta’s new products. The company also announced that it will be rolling out AI chatbots across its family of apps, in an effort to boost engagement.

(2) Artificial Intelligence Could Finally Let Us Talk with Animals

AI is not only breaking down human language barriers, it might also make understanding how animals communicate possible. The article linked above profiles a number of projects and milestones from research teams using AI to decode animal communications. One team, for example, is planning on mic’ing up large swathes of the ocean to train a model to speak whale. THATS SO COOL BRO

Of course, decoding animal communications behaviors won’t automatically give animals human-level consciousness and intellect. A bakri will be a bakri only lol. But, the translations could give us a better understanding of how and why animals behave in the peculiar ways that they do, which in turn, could help us help them.

And hopefully, translation will go both ways, imma ask this dog why is he SUCH A MOOD always

(3) Microsoft is going nuclear to power its AI ambitions

Justine Calma for The Verge

All of the AI work we profiled above is going to take a lot of energy. Microsoft is, apparently, looking towards nuclear to power some of it. The company revealed that it’s hiring someone to “lead project initiatives for all aspects of nuclear energy infrastructure for global growth.” Microsoft, like almost all major tech companies, has made clean energy commitments — and keeping those promises while also keeping the lights on at a growing data center footprint is going to require a full menu of solar, wind, advanced geothermal, and nuclear.

Also one of the reports from the AVP, at XelpMoc

#CleanCompute as an urgent problem is something we've been thinking about while To illustrate, a single training run of GPT-3 releases 93,000kg of Co2 equivalents in France. That same training run is a staggering 858,360 Kgs in India - almost 10x!
Water usage to run these models is also skyrocketing - Presenting new and important challenges to big Tech.

Additionally, Training large ML models used to be the most resource-intense component of setting up and delivering AI. In the world of Large Language Models, We see that not just training, but running inferences also are increasingly GPU intensive.

This important problem poses multiple opportunities for startups with huge and growing markets- there are several million data centres globally, Each power-hungry facility can pack in tens of thousands of servers, guzzling environmental resources more than entire countries.

I look at this as an opportunity for us to disrupt further, data is the new oil? it stinks of pollution friends.

See you next week, Enjoy your weekend