OpenAI delivered on its promise of “magic” at its Spring Update event today, unveiling the highly anticipated GPT-4o model that brings multimodal capabilities to both the free and paid versions of ChatGPT. This powerful new AI can understand and generate speech, analyze images and videos, and even power a more natural and emotional-sounding voice assistant.
While OpenAI kept some of its cards close to the vest, like details on the next-gen GPT-5 model and the release of its AI video generator Sora, the GPT-4o reveal packed more than enough punch to leave the AI-hungry audience buzzing.
Free ChatGPT Gets Custom Chatbots
One of the most exciting announcements for free ChatGPT users is that they’ll soon be able to create custom chatbots powered by GPT-4o. This allows broader access to OpenAI’s latest language model capabilities.
More Efficient, Multimodal GPT-4o
At the core of today’s news is GPT-4o, a more efficient multimodal AI model that will drive both the free and paid ChatGPT experiences. By being multimodal by design, GPT-4o can seamlessly analyze inputs like images, videos, and speech and respond accordingly in text or via speech.
Human-Like ChatGPT Voice
Leveraging GPT-4o’s speech capabilities, OpenAI unveiled ChatGPT Voice – a stunningly natural-sounding AI voice assistant. Demos showed ChatGPT Voice exhibiting more emotional range and human-like inflections than existing AI assistants.
Multimodal ChatGPT Desktop App
Further immersing users, OpenAI is launching a ChatGPT Desktop app that combines the AI’s text, voice, and vision capabilities into one unified experience.
Live Translation Game-Changer
One GPT-4o feature that could revolutionize global communication is real-time voice translation. Like human interpreters at international events, GPT-4o can translate back-and-forth between speakers of different languages. While imperfect, its ability to allow live, interruptible voice conversations is groundbreaking for travelers and businesses.
“Paradigm Shift” in Human-Computer Interaction
As OpenAI’s CTO Mira Murati commented, the ability to simply speak to an AI assistant that deeply understands multimedia inputs feels like a “paradigm shift” in how we interact with computers and data. While certain details like precise release dates remain unclear, one thing is certain – with GPT-4o, the “magic” of AI has become multimedia. OpenAI has raised the bar for what We can expect from human-computer interaction.
FAQ:
What enhancements does GPT-4o bring?
In the update, GPT-4o has the ability to mimic human cadences in its verbal responses and can even attempt to detect people’s moods. This feature draws parallels to the 2013 film “Her,” where the protagonist develops a relationship with an AI system.
What are the changes specific to ChatGPT in the new version?
OpenAI claims that the updated version operates faster than its predecessors and can engage in reasoning across text, audio, and video in real-time. GPT-4o will be the engine behind OpenAI’s widely-used ChatGPT chatbot.
Who can access the updated version?
GPT-4o will be rolled out to all users in the coming weeks, including those who utilize the free version of ChatGPT, according to OpenAI’s announcement.
What are experts saying about GPT-4o?
“By combining voice, text and images seamlessly, the demo captured how AI can drive incredible amounts of productivity,” said Bank of America. “Much of the demo used a ‘hardwired’ iPhone to lower latency of real time interaction.” Gartner analyst Chirag Dekate noted that the update indicates OpenAI is catching up to larger competitors. He mentioned similarities between OpenAI’s demonstrations and capabilities with those showcased by Google in their Gemini 1.5 pro launch. Dekate highlighted emerging capability gaps compared to peers, particularly Google, despite OpenAI’s initial advantage with ChatGPT and GPT-3.
