ChatGPT will soon get speech and vision

The interface that made her famousArtificial intelligenceAmnesty International Generator (capable of producing text, images and other content based on a simple request in everyday language) will soon be able to process requests containing images as well as discuss verbally with its users.

Internet users will be able, for example, to take a photo of a monument and Have a chat with ChatGPT on the building’s history, or even show the program what’s in their refrigerator so it can give them a recipe, OpenAI suggests in a press release.

These new tools will be rolled out over the next two weeks to paid members of the ChatGPT Plus service, or to organizations that are customers of the service.

The company announced the anticipated addition of such features last March, when it introduced GPT-4, the latest version of its language model. (A new window)

GPT-4 is multimedia, meaning it can handle data other than text or computer code.

dangers From hallucinations It still exists

The success of ChatGPT since late 2022 has led to a big race forArtificial intelligenceAmnesty International Generative between technology giants Google and Microsoft in the lead.

However, the rapid deployment of these still poorly organized programs also raises a lot of concern, especially as it tends to hallucinationThis means inventing answers from scratch.

The models that support the vision present new challenges, introduces OpenAI in a press release. Among these, the company notes Hallucinations they may have, but also the risk of trusting the model’s interpretation of images in high-risk areas.

The startup claims to have Model testing On topics such as extremism and scientific knowledge, it relies on real-life uses and netizen feedback for improvement.

It also limits ChatGPT’s capabilities People analysisBecause the interface It is not always accurate and these systems must respect the confidentiality of individuals.

Spotify partners with OpenAI

On Monday, streaming platform Spotify also announced a partnership with OpenAI to translate podcasts liveArtificial intelligenceAmnesty International.

Presentations recorded in English will now be presented in other languages While retaining the distinctive acoustic characteristics of the speakerThe service said in a statement.

The Swedish company confirms that the new audio generation technology is from OpenAI Reproduces the original speaker pattern, allowing for a more authentic, personal and natural listening experience than traditional dubbing.