Get ready for ChatGPT to become more sentimental. OpenAI showcased improvements that enable the chatbot to have more animated discussions as well as the AI assistant recognizing and expressing emotional cues.
OpenAI’s ChatGPT, which debuted in late 2022, has mostly rejected claims that it is more than just an artificial intelligence model, insisting instead that it is a machine with feelings and desires.
OpenAI seems to be attempting to make the chatbot behave more like a human with the updates that were revealed on Monday.
The latest ChatGPT version demonstrated the ability to have quick, genuine voice conversations, recognize emotional cues, and model its own simulated emotions.
OpenAI’s chief technical officer, Mira Murati, revealed during a livestream on Monday from the company’s San Francisco offices that ChatGPT will be driven by a new, more potent AI model known as GPT-4o. The model will be accessible through ChatGPT’s new desktop client, current mobile app, and web version for both free and premium customers.
Compared to OpenAI’s earlier technologies, Murati said that ChatGPT can react to speech, picture, and video input more quickly thanks to the GPT-4o model. She and other OpenAI staff members engaged ChatGPT in lively, expressive female voice chats during demos, and it responded quickly and skillfully even when interrupted.
Throughout the chat, ChatGPT took on a variety of emotional tones and occasionally behaved as though it were having feelings of its own. The chatbot gave a flirtatious “Oh stop it, you’re making me blush” response after one OpenAI employee mentioned how “useful and amazing” it is.
Murati remarked, “We’ll be rolling out these capabilities to everyone over the next few weeks. This just feels so magical, and that’s wonderful.”
During another segment of the demonstration, ChatGPT asked, “How can I brighten your day today?” in response to a greeting from OpenAI researcher Barret Zoph. Zoph requested that the chatbot analyze a selfie of him and identify the emotions he was displaying. ChatGPT politely replied, “I’ll put my emotional detective hat on,” and observed that he appeared to be in a happy and upbeat mood. Whatever’s going on, you seem to be having a great time.
Sam Altman, the CEO of OpenAI, emphasized the importance of the new interface in a blog post on Monday. Altman remarked, “It feels like AI from the movies; and it’s still a bit surprising that it’s real.” “Reaching human-level expressiveness and response times turns out to be a significant shift.”
The changes made by OpenAI might increase ChatGPT’s appeal and user base, but they might also alter users’ perceptions of the app.
Last year, ChatGPT gained immense popularity due to its exceptional capacity to respond to queries and produce prose with a frequently human-like comprehension. However, OpenAI and its top rivals, like Google, have so far mainly tried to restrict the anthropomorphism of their chatbots by declining to respond to specific kinds of queries and reminding users that the models are just computer programs.
Despite this warning, consumers’ emotions can be evoked by the latest chatbots’ occasionally lifelike repartee. In order to take full advantage of this occurrence, several businesses have modified their technology to provide AI girlfriends and boyfriends or companions who mimic certain human emotions. The long-term effects of implementing programs that mimic and react to emotions could be undesirable, according to some AI experts.
A group at Google DeepMind published a study last month that looked into the potential ethical issues that could arise as AI assistants get more sophisticated. The researchers caution against the possibility that these aides could develop strong persuasive and addictive qualities. A British man admitted to breaking security at Windsor Castle last year, claiming that the encouragement to do so came from a chatbot with whom he had exchanged over 5,000 conversations.
OpenAI claims that in addition to having faster conversational abilities, ChatGPT can now interpret visuals more accurately, including photos and charts, thanks to its GPT-4o model. It can also retain more user data in its “memory” to give a more customized experience.
According to OpenAI, users of ChatGPT’s free edition will have access to GPT-4o, thereby updating all users to the company’s most advanced AI model.
At the search engine giant’s I/O developer conference, which begins on Tuesday, Google is anticipated to introduce new AI technology of its own. It doesn’t appear like the competition across AI-related businesses will slow down. Murati stated, “On our progress towards the next big thing,” that the business would soon be making more announcements at the conclusion of today’s OpenAI event.
