Welcome to this week’s roundup of handwritten AI news.

This week OpenAI said its AI voice cloner was too good to release.

An NYC-approved chatbot gave residents dodgy legal advice.

And researchers built an AI that knows what’s going to occur in the long run.

Let’s dig in.

AI and the feds

AI is ready for deeper integration in US government operations. The White House announced latest AI rules for federal agencies to make sure they use the tech in a way that’s secure and doesn’t compromise the rights of the people they serve.

The US government may even launch a giant hiring drive to employ 100 “AI professionals” to deploy to federal agencies. Good luck finding AI talent that hasn’t already been snapped up by Meta et al.

In a worldwide push for AI safety, US and UK ministers met to ascertain a bilateral agreement on AI safety. The almost complete absence of UK AI regulation might even see the US take the lead in forming critical policy.

The agreement says the countries will cooperate and share research in some interesting areas.

If they don’t manage to agree on the way to keep us secure from AI, let’s hope a glitch within the training data saves us.

Source: Pinterest

I’m pretty sure it’s legal

If you’re searching for a very good reason why the deployment of AI in government agencies needs regulation then look no further than New York City. The city launched an AI chatbot to assist residents and businesses navigate government regulations and it’s advising people to interrupt the law.

The NYC Office of Technology and Innovation says people should chill because most of what the chatbot says is true and besides, it’s a “work in progress.”

Google’s DeepMind may give you the chance to assist with issues like this. It developed SAFE, an AI agent that may fact-check LLMs. For now, SAFE just gives us an interesting solution to benchmark which AI model delivers probably the most factual long-form answers.

Could it will definitely help cure AI’s tendency to make stuff up?

DeepMind also developed Genie, a foundational model for constructing 2D game environments from images or text prompts.

The researchers assure us their text-to-game model isn’t only a solution to get kids to make and play more computer games. It has interesting applications for robotics too.

Where do I enroll for it? I intend to make use of Genie purely for robotics research. And perhaps to recreate my version of Mario Brothers.

Prime AI

The AI model leaderboard is hotly contested, but there’s little doubt which model Amazon thinks will win. The company just invested one other $2.75 billion into Anthropic, the makers of the Claude models.

Apple is yet to release its own LLM, but there was a flurry of AI developments at the corporate after what appeared to be a slow start. Apple researchers developed ReALM, a model that ‘sees’ on-screen visuals higher than GPT-4.

Could Siri get an AI upgrade soon?

Who said that?

AI voice clones are already good and keep recuperating. OpenAI gave us a couple of samples of how good Voice Engine, its voice cloning text-to-speech model is.

The company says initial testing shows there are some great use cases for the tech. However, because Voice Engine is so good the corporate says it could be too dangerous to release.

Speaking of dangerous behavior, is a relationship with an AI companion roughly dangerous than the true thing?

AI relationships appear to be on the rise. Want to know which countries are most inquisitive about AI girlfriends? Here’s a full rundown of world AI girlfriend trends. You could also be surprised who likes the thought and who doesn’t.

Humans are weird, so it’s unsurprising that training an AI to act how we would like it to is a challenge. A brand new study attempted to align AI with crowdsourced human values. Is that a very good thing?

Imagine if crowdsourced values were used to coach an AI in Germany in 1939, or at the peak of the slave trade.

Our faith in humanity was tested after we forgot to examine the calendar before getting excited over this X post from Hugging Face engineer Phillip Schmid.

Casual Easter Monday with an enormous gift from @OpenAI!🤯 They just released an old GPT-3.5 version. 😍

👉 https://t.co/wo5VshOSfC pic.twitter.com/IPXbnd8IoH

— Philipp Schmid (@_philschmid) April 1, 2024

Sam must have pranked him by just releasing it.

Future power

Power usage within the AI industry continues to attract attention for all of the improper reasons. At NVIDIA’s GTC event, Delta Electronics unveiled latest energy-efficient AI hardware that might help with that.

It may not grab headlines just like the Blackwell GPUs, however the tech is impressive and so is the quantity of power it should save.

What does the long run of AI hold? Who will win the US election? Will humans go to Mars before 2030?

We might get one of the best answer by asking GPT-4. Berkeley researchers built an AI forecasting system that’s more accurate than humans. It does have an odd quirk though.

In other news…

Here are another clickworthy AI stories we enjoyed this week:

  • Microsoft and OpenAI are discussing a partnership to construct the $100B Stargate supercomputer.
  • Google desires to run Gemini on its Pixel 8 phones but says it should require at the least 8GB of RAM.
  • Elon Musk estimates there’s a 20% probability AI will destroy humanity, but that shouldn’t stop AI development. His was one in all the more optimistic estimates.
  • NYC to check AI gun detection on the subway. Previous tests of the tech didn’t go thoroughly.
  • OpenAI says you should use ChatGPT without signing up now. But there’s a catch.
  • Opera allows users to download and use LLMs locally.
  • Did Emad Mostaque leave Stability AI over unpaid bills?

Stability AI reportedly ran out of money to pay its bills for rented cloudy GPUs https://t.co/IuGvQ9SCf5

— The Register (@TheRegister) April 3, 2024

And that’s a wrap.

Let’s hope AI gets rolled out faster in US federal agencies. If you’ve ever needed to cope with the DMV or the Labor Department then it could only improve. Right?

Groundbreaking AI developments are being made day by day, however the tech I really need to play with is DeepMind’s Genie. I would like to make use of AI to make 2D platform games like we had back within the 80’s. Is that just a little weird? Who’s with me on this?

AI girlfriends; a very good idea for lonely people or simply weird? I’m considering a digital twin of your companion could assist you to run simulations that might avoid loads of arguments before they begin.

Let us know what you think that, and send us links to any juicy AI stories we could have missed.

The post DAI#33 – Games, voice clones, and AI fortune tellers appeared first on DailyAI.


This article was originally published at dailyai.com