The timing was good: exactly one day before the start of Google’s developer conference I/O, the newly growing competitor OpenAI presented a new version of its AI application Chat-GPT on Monday evening German time. Chat-GPT0, the new Large Language Model from the Californians, unlike its predecessors, combines speech recognition, processing (what is meant) and the output of the answer in spoken language with image recognition, and does so much faster than before. According to head of technology Mira Murati, this increased efficiency allows the new version to be offered free of charge worldwide. OpenAI is setting the bar high again.
Chat-GPTo therefore responds significantly faster to inquiries and can also recognize emotions in faces or voices. The voice output also knows different emotional states. If you previously had to let Chat-GPT finish, you can now interrupt the system.
In some live demos, Murati’s employees showed off some of the new capabilities. Among other things, Chat-GPT0 helped solve a simple linear equation that an employee had written on a piece of paper. If requested, the friendly voice gave tips on what could possibly be done next. In another demo, the system explained what a piece of computer code was about. And finally, it translated a few bits of Italian into English live without much delay.
When it comes to security, things get murky
Murati also announced a version of Chat-GPT 4o for desktop computers. As Murati said, it has been optimized for 50 languages. Paying users get the new version 50 percent cheaper than before and can survey it five times as often as non-paying users.
Murati remained rather vague on the subject of security. Yes, the new abilities create dangers, for example through the lifelike speech output. A lot of work has been done to make the system as secure as possible and we are in contact with the authorities and responsible institutions.
The presentation is probably not a shock for the competition, but it is significant enough. A huge material battle has broken out among the big tech companies over who can secure dominance in the Large Language Models. In the medium term, it will no longer just be about language and computer code, but also about the building blocks of life and new materials, for example for batteries.
And with OpenAI, Murati hinted, there could soon be more news; the “next big thing” will soon be presented.