On Monday, OpenAI conducted a live demo of GPT-40, its newest large language model (LLM), as the generative AI (GenAI) arms race ramped up.
The California-headquartered tech company showcased how its next-generation LLM enables the ChatGPT chatbot to engage in real-time, spoken conversations. OpenAI Chief Technology Officer Mira Murati described the model as proof of the firm’s commitment to maximizing ease of use.
Executives illustrated the LLM giving spoken instructions on how to solve a math problem and a coding issue. Moreover, they showed it telling a bedtime story, even singing certain parts in a natural, human-sounding voice.
OpenAI’s event came a day before the annual Google developer conference, which is expected to reveal updates to Gemini AI. Like the GPT-40, Gemini is a multimodal model that can interpret and generate audio, image, and text.
In addition, the AI model demonstration preempted Apple’s Worldwide Developers Conference, which is scheduled for next month. Tech specialists expect the summit to focus on new ways of integrating GenAI into the next iOS or iPhone releases.
Industry watchers anticipate Microsoft, which has invested as much as $13.00 billion in OpenAI, as the biggest beneficiary of GPT-40. They added that the market may see voice generation technology embedded in Microsoft’s next batch of new product offerings.
GPT-40 Prone to Scammers, Needs OpenAI Protections
New Zealand-based online safety organization Netsafe warned that the OpenAI GPT-40 can be a potent scamming tool. It only needs 15 seconds of clear recording to recreate a voice and make it say anything the user wants.
The non-profit stressed that an AI-generated phone call is more likely to convince the victim than a text message. It expects scammers to target senior citizens, asking for money using copied voices of a target’s children or grandchildren.
Lastly, OpenAI acknowledged the potential risks GPT-40 could bring and reassured that it was developing the necessary countermeasures. The AI creation and deployment firm added that it delayed the new LLM’s release to ensure enough safeguards.
