- The Daily Update
- Posts
- ChatGPT's new humanlike Voice Mode
ChatGPT's new humanlike Voice Mode
OpenAI launches GPT-4o
Welcome back!
We’re back with huge news from OpenAI. The company just launched a new model that’s redefining human-computer interactions. Let’s go.
In today’s Daily Update:
🗞️ OpenAI launches ChatGPT-4o
🤖 US ramps up military AI race with China
📸 Apple nears deal to bring ChatGPT to iPhones
🚨 AI Roundup: Four quick hits
Read time: 3 minutes
TOP STORY
🗞️ OpenAI launches ChatGPT-4o
Yesterday OpenAI launched an updated version of ChatGPT that features industry-leading reasoning across audio, visuals and text. The model achieves the most humanlike interactions of any GPT to date.
What you need to know:
GPT-4o (“omni”) was trained end-to-end across all three input types. This means that instead of converting text to other outputs, the model produces audio and visual responses in real time.
Omni’s Voice Mode responds to audio inputs in 320 milliseconds on average, which is similar to human response times in conversations.
GPT-4o appears to be the world’s most advanced AI model across traditional benchmark tests.
The model will soon be available to all ChatGPT users, although paid users will have 5x higher message limits.
Why it matters: I received access to GPT-4o yesterday, and it certainly lives up to OpenAI’s claims. This is the most advanced AI model in the world and OpenAI is just scratching the surface of its new multimodal training technique. Generative AI is evolving rapidly, and we’re getting close to models that can serve as useful personal assistants for everyone.
Food for thought: GPT-4o’s Voice Mode is incredibly humanlike, but one of the most shocking demo videos shows two different GPTs interacting with each other. The GPTs quickly start a fairly natural conversation about the room around them, and eventually figure out how to sing a duet together when prompted by OpenAI President Greg Brockman. Experts say that AIs interacting with each other is incredibly dangerous, but it looks like we’ve already reached a point where that’s possible.
AI INSIGHT
🤖 US ramps up military AI race with China
DALL-E 3
The U.S. military is ramping up efforts to stay ahead of China’s AI weapon systems despite public concerns.
Key points:
The Air Force recently tested an AI-piloted fighter jet in a dogfight against a human pilot.
There are currently hundreds of other ongoing AI projects across the Pentagon.
U.S. officials say that future wars will never be fought by machines without direct human oversight, but questions linger about China’s stance on the issue.
The U.S. military sees the race for AI supremacy as the only way to mitigate risks associated with China’s use of the technology.
Why it matters: The future of warfare is starting to look very interesting (and potentially concerning). AI weapons systems are advancing at unprecedented speeds and neither the U.S. nor China aims to slow down.
BUSINESS SPOTLIGHT
📸 Apple nears deal to bring ChatGPT to iPhones
DALL-E 3
Apple is reportedly closing in on a deal with OpenAI to bring ChatGPT-powered features to the iPhone.
The details:
The two companies are finalizing terms to include ChatGPT in Apple’s iOS 18.
This agreement would allow Apple to offer an advanced chatbot on the next generation of its devices.
Apple CEO Tim Cook personally uses ChatGPT, but said last year that there are “a number of issues that need to be sorted.”
Cook also believes that Apple will have an edge in AI, but could be referring to the company’s ability to seamlessly integrate new services into its devices.
Why it matters: Bloomberg’s Mark Gurman is normally on the money with all news Apple-related. Given the impressively humanlike capabilities of GPT-4o, Apple’s Siri 2.0 is shaping up to be one of the most advanced AI assistants in the world.
MORE TRENDING NEWS
🚨 AI Roundup: Four quick hits
Source: OpenAI
OpenAI launches ChatGPT desktop app for macOS and makes the GPT Store free to all users.
Wikipedia creates experimental Chrome extension that fact-checks websites with ChatGPT.
Anthropic expands Claude chatbot to Europe with support for French, German, Italian and Spanish.
Russian-linked networks use AI to rewrite real news stories and spread propaganda.
THAT’S ALL FOR TODAY