Satya Nadella, CEO of Microsoft Corp., throughout the firm’s Ignite Highlight occasion in Seoul, Nov. 15, 2022.
SeongJoon Cho | Bloomberg | Getty Photographs
Microsoft on Tuesday introduced a big replace to its synthetic intelligence chatbot: visible search. Customers can now take or add a photograph to Bing Chat and ask for extra info on it by means of the desktop or Bing app.
“Bing can perceive the context of a picture, interpret it, and reply questions on it,” Microsoft wrote in a launch. “Whether or not you are touring to a brand new metropolis on trip and asking in regards to the structure of a selected constructing or at house attempting to give you lunch concepts primarily based on the contents of your fridge, add the picture into Bing Chat and use it to harness the net’s data to get you solutions.”
The replace comes because the AI arms race heats up amongst chatbot leaders comparable to Microsoft, Google, OpenAI and Anthropic. Within the effort to develop essentially the most superior generative AI, tech giants are shortly launching new options, aiming to maintain up with not solely their text-based chatbot opponents, but in addition image-heavy AI instruments.
Though picture search and responses that embrace pictures at the moment are turning into a part of the person expertise for chatbots, not one of the main text-based chatbots appear to have the ability to generate their very own pictures but, not like instruments comparable to Midjourney, Secure Diffusion and DALL-E 2. Nonetheless, Google says the characteristic is on the best way for its Bard chatbot.
Microsoft’s determination to permit pictures for Bing Chat follows Google’s recent debut of a picture search characteristic for Bard, its AI chatbot. Utilizing Google Lens, customers can request info from Bard about a picture they’ve uploaded, ask it to generate a caption and even simply add some zest to the chatbot’s responses, comparable to a request for restaurant suggestions with images of the restaurant’s interiors included. On the time of writing, OpenAI’s ChatGPT doesn’t permit picture uploads, because the chatbot remains to be fully text-based, and Anthropic’s chatbot, Claude 2, operates equally.