Google AI and DeepMind News and Discussions

[wjfox]

[weatheriscool]

Google Bard Updated With Text to Speech, 40 New Languages
Bard is getting more capable, but it's still not trustworthy.
By Ryan Whitwam July 14, 2023
https://www.extremetech.com/extreme/goo ... -languages

Google was caught off-guard earlier this year when Microsoft decided to make generative AI its new focus. Despite inventing the transformer algorithms that make bots like ChatGPT possible, Google's answer to ChatGPT stumbled out of the gate. Google has been updating its Bard AI consistently in recent months, and the latest update is a big one. Finally, Bard can speak its replies.

Bard is a text-based generative AI, which means you can ask it anything, and it'll give you a response. It can do that because Bard has ingested a huge amount of written content, so it has the uncanny ability to generate text that sounds like a (boring) human being wrote it. On the flip side, some of the text Bard creates is not grounded in reality. So far, no one has figured out how to prevent these "hallucinations" in generative AI—ChatGPT suffers from the same shortcoming.

Google says it has now fed Bard enough content in an assortment of languages to expand the bot's services. It now works in 40 new languages, including Arabic, Chinese, German, and Spanish. As part of the multilinguistic update, Bard has also expanded to Brazil and more of Europe.

[spryfusion]

[spryfusion]

Project page with demos: https://robotics-transformer2.github.io/
NYT article: https://archive.ph/objpJ

[Powers]

Project page with demos: https://robotics-transformer2.github.io/
NYT article: https://archive.ph/objpJ

It learns from both web

That will not end well.

[weatheriscool]

Google's Gemini is so powerful that OpenAI probably won't release GPT-5 until late 2025
AI

Google announced that it will release Gemini in a few months. If the information presented in this recent YouTube video is correct, OpenAI, Anthropic, Meta, Microsoft, Stability, xAI, Apple, Amazon and every other would be competitor will have to come up with some very unexpected major breakthroughs to keep pace.



In fact, to compete with Google the other AI developers may have no choice but to form collaborations of no fewer than three of them. That's how powerful Gemini promises to be.

Perhaps the most important part of Google's new Gemini series will be its AI medical advisor. Who would have thought that we are so close to everyone having a doctor on their smartphone that is far more intelligent and knowledgeable than the most intelligent and knowledgeable human doctor today? Google has just revolutionized healthcare in a major way.

The other part of Google's power comes from it's multi-model approach. Be prepared for Gemini to be able to easily out-think every other model, and perhaps be approaching AGI much sooner than anyone would have expected.

[spryfusion]

Google nears release of AI software Gemini, The Information reports
Source: Reuters

Google has given a small group of companies access to an early version of Gemini, its conversational artificial intelligence software, The Information reported on Thursday, citing people familiar with the matter.

Gemini is intended to compete with OpenAI's GPT-4 model, according to the report.

For Google, the stakes of Gemini's launch are high. Google has intensified investments in generative AI this year as it plays catch-up after Microsoft-backed OpenAI's launch of ChatGPT last year took the tech world by storm.

Gemini is a collection of large-language models that power everything from chatbots to features that either summarize text or generate original text based on what users want to read like email drafts, music lyrics, or news stories, the report said.

It is also expected to help software engineers write code and generate original images based on what users ask to see.

Google is currently giving developers access to a relatively large version of Gemini, but not the largest version it is developing which would be more on par with GPT-4, the report said.

The search and advertising giant plans to make Gemini available to companies through its Google Cloud Vertex AI service.

[wjfox]

Google DeepMind AI speeds up search for disease genes

29 minutes ago

Google's AI firm DeepMind has used artificial intelligence to identify changes in human DNA that might cause diseases.

The researchers believe they have pinpointed 89% of all the key mutations.

The development is expected to speed up diagnosis and help in the search for better treatments.

A leading independent scientist told BBC News that the work was "a big step forward".

Prof Ewan Birney, deputy director general of the European Molecular Biology Laboratory, said: "It will help clinical researchers prioritise where to look to find areas that could cause disease."

https://www.bbc.co.uk/news/science-environment-66847977

[raklian]

DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.

DeepMind cofounder Mustafa Suleyman wants to build a chatbot that does a whole lot more than chat. In a recent conversation I had with him, he told me that generative AI is just a phase. What’s next is interactive AI: bots that can carry out tasks you set for them by calling on other software and other people to get stuff done.

The first wave of AI was about classification. Deep learning showed that we can train a computer to classify various types of input data: images, video, audio, language. Now we’re in the generative wave, where you take that input data and produce new data.

The third wave will be the interactive phase. That’s why I’ve bet for a long time that conversation is the future interface. You know, instead of just clicking on buttons and typing, you’re going to talk to your AI.

And these AIs will be able to take actions. You will just give it a general, high-level goal and it will use all the tools it has to act on that. They’ll talk to other people, talk to other AIs.

https://www.technologyreview.com/2023/0 ... st-a-phase

[Cyber_Rebel]

I guess things like recursive self-improvement. You wouldn’t want to let your little AI go off and update its own code without you having oversight. Maybe that should even be a licensed activity—you know, just like for handling anthrax or nuclear materials.

Or, like, we have not allowed drones in any public spaces, right? It’s a licensed activity. You can't fly them wherever you want, because they present a threat to people’s privacy.

But who's to say that I wouldn't want my whittle AI to go off an update its own code? Perhaps I need my whittle AI to be more capable in its task, and the best way to efficiently make that happen is for it to recursively self-improve? Isn't that not also an invasion of my privacy?

How far are we even truly for AI being able to do that?

But you need top-down regulation too. I love the nation-state. I believe in the public interest, I believe in the good of tax and redistribution, I believe in the power of regulation. And what I’m calling for is action on the part of the nation-state to sort its shit out. Given what’s at stake, now is the time to get moving.

Dude's brilliant, not going to question that, but I feel like a lot of the assumptions he tends to make are based on this ideology, where we have to somehow control AI to keep business as usual. It just seems hypocritical to call on the nation-state to control something which he essentially helped create in the first place, for control rather than for a new framework entirely. AI is very different to all of the regulatory examples he's mentioned.