• Starter AI
  • Posts
  • Apple’s ReALM, US & UK for AI safety, OpenAI’s Voice Engine

Apple’s ReALM, US & UK for AI safety, OpenAI’s Voice Engine

Two countries are putting efforts to regulate AI.

Hello, Starters!

In AI, being first is something very important: the first to create, the first to develop, the first to regulate. Europe made this step recently; now the US seems to be following that path.

Here’s what you’ll find today:

  • Apple researchers unveil “ReALM”

  • US & UK reach the world’s first bilateral AI deal

  • OpenAI’s approach to Voice Engine

  • ChatGPT requires no sign up

  • DeepMind’s Demis Hassabis warns against AI hype

  • And more.

Apple continues to move steadily through the AI frenzy with the development of a system dubbed ReALM (Reference Resolution As Language Modeling), designed to enhance the performance of voice assistants like Siri.

ReALM is capable of catching up with the user's activity through screen-based reference. In other words, besides understanding complex tasks, it can also use what's happening on the user's screen for context and to provide better results, harnessing Apple's in-house and yet-to-be-revealed AI models, which showcase capabilities that could surpass GPT-4.

As more advanced AI models are expected to emerge next year, the US and UK have taken a groundbreaking step by reaching an agreement on AI safety. This agreement will facilitate the exchange of expertise and collaboration in evaluating private AI models.

The deal represents the first formal cooperation between countries to address the potential risks of AI and was signed in Washington by UK Science Minister Michelle Donelan and US Commerce Secretary Gina Raimondo, highlighting the ongoing efforts of governments to establish regulations and promote the development of responsible AI technologies.

OpenAI's Voice Engine has been gaining widespread attention recently, ever since we first learned about it through a trademark filing. Now that the company has disclosed how they're approaching its "launching," it's worth reflecting on how synthetic voice will impact our reality.

That's the main reason behind OpenAI's decision not to release this technology yet widely. The company remains committed to developing safe and beneficial AI. They emphasise that despite the benefits Voice Engine could bring to reading assistance, translation, and aiding people with disabilities, there are also serious risks associated with it.

Nonetheless, it's a powerful example of AI's potential and one of the many groundbreaking applications we're yet to see in the future.

✅As of now, OpenAI's ChatGPT no longer requires users to sign up to start asking queries. This comes as a move to reach a wider audience. However, the feature is rolling out gradually and doesn't include chat history or the option to not train with your data.

💭Google DeepMind founder Demis Hassabis recently shared his thoughts on AI's impact, claiming that the current hype is similar to what we once saw with cryptocurrency. The CEO states that this focus is damaging to other relevant AI developments such as scientific discovery.

What did you think of today's newsletter?

Login or Subscribe to participate in polls.

Thank you for reading!