Tue 19 November, 2024 - AI Omegle

WELCOME

Hey all, welcome back to The Frontier — our weekly newsletter covering the hottest new launches in AI and industry trends. This week, we’ve got five killer AI apps for you to try out and we’re diving into whether or not AGI is around the corner.

TOP LAUNCHES

An AI-powered Omegle, a better way to keep up with the news, and more.

Top launches

Friend: A platform that connects you to random AI "friends."

Friend, the company that made waves on tech Twitter for spending $1.8M on its domain, just launched what feels like an AI version of Omegle. It connects you to random "friends," each with their own baggage, for one-on-one chats.

Knowing: Interact with LLMs inside nested concept hierarchies.

Knowing offers a much more intuitive way to gather and store AI-generated knowledge than the typical prompt-response UI. It lets you create nested concept hierarchies (e.g. “Philosophers -> 18th Century -> German -> Immanuel Kant”) and populate each node with AI-generated information you can prompt in the app.

Projects by ElevenLabs: Structure, edit, and generate long-form audio with precision.

Projects is ElevenLab's new end-to-end workflow that lets you upload very long documents and turn them into audio. Essentially, it gives you everything you need to create high-quality audiobooks on demand in minds 

Particle: An easier way to keep up with the news. 

Particle is a news app that’s built to help you stay up to date without burning out. It breaks down complex stories into concise summaries so you can spend less time trying to get the facts. It also uses AI to allow you to ask further questions for each story if you feel some context might be missing. 

Output Media API: API to build interactive AI agents that talk in meetings.

Recall.ai’s Output Media API lets you build AI agents that join Zoom, Meet, Teams, and Webex video conferences. These agents can output audio and video, listen, and respond like real participants. Are we set for a world of no human meetings? Maybe. 

Next-Level Speech AI

Universal-2 sets a new standard in Speech AI, delivering precise, source-of-truth transcripts that capture the true complexity of human speech. Turn audio into clear, accurate text with advanced speech-to-text capabilities that go beyond traditional metrics. Try for free with $50 in API credits.

THE BIG IDEA

AGI when, chat?

AGI in 2025? That’s what Sam Altman seems to think. The OpenAI CEO recently claimed that Artificial General Intelligence—the kind of AI that can out-think you at chess and then help you write a novel about losing—is just around the corner. Altman says the roadmap is “basically clear,” and that we already have the hardware needed to make it happen. Just a matter of tweaking and tuning, apparently.

Dario Amodei, CEO of Anthropic, is slightly less bullish but still optimistic. He thinks AGI could show up by 2026 or 2027, assuming we navigate a few hiccups like data shortages and, you know, global politics. But it’s not all champagne and existential dread in the AI world. Some reports suggest OpenAI’s next big model, Orion, isn’t the game-changer many hoped for—more evolution than revolution. Scaling current models might not be the golden ticket to AGI after all.

And then there’s the practical stuff: the immense cost of training these models, the environmental toll, and the logistical headache of powering it all. Big tech is doubling down on infrastructure, but the sustainability of this arms race remains a giant question mark.

So, are we on the brink of a sci-fi future where AGI redefines everything, or are we just hyping up what could be a glorified calculator?

Overheard in the discourse

OpenAI’s email archive

As part of the ongoing court case between Elon Musk and OpenAI, a bunch of internal emails between Elon, Sam Altman, Ilya Sutskever, and Greg Brockman have been released to the public. They’re a fascinating read — check them out here

Reply

or to participate.