AI chat tool ChatGPT is getting new features that will allow its users to have a voice conversation and share images to explain problems or ask for advice.
OpenAI made the announcement this week and revealed that the new functionality will roll out to ChatGPT Plus and Enterprise users within the next two weeks. Voice and chat will reach other users shortly after.
The voice functionality basically does what it says: you’ll be able to ask ChatGPT a question and have it respond in what OpenAI calls a “back-and-forth conversation.”
The feature uses an audio-based text-to-speech model from five voice actors, and you can choose the voice you’re most comfortable with.
However, using text-to-speech in this way is not without drawbacks: OpenAI is well aware of the ability to create voice deepfakes.
“That’s why we’re using this technology to power a specific use case: voice chat,” OpenAI said in a statement. “Voice chat was created with voice actors we’ve worked with directly.”
OpenAI is also working with Spotify so podcasters can use the same technology for language translation.
The image input is designed to add another dimension to the ChatGPT query. As an example, OpenAI says you’ll be able to take a photo of your refrigerator and ChatGPT will be able to tell you what you can do with what’s there, as well as how to do it.
Again, OpenAI is aware of some of the problems such a feature could have.
“Prior to broader deployment, we tested the model with red team members for risks in domains such as extremism and scientific competence, and a diverse set of alpha testers,” OpenAI said. “Our research allowed us to align on some key details for responsible use.”
According to OpenAI, the key to these AI challenges is to roll out features slowly, allowing the company to “make improvements and refine risk mitigation over time while preparing everyone for more powerful systems in the future.”
“We have also taken technical steps to significantly limit ChatGPT’s ability to analyze and make direct statements about people, as ChatGPT is not always accurate and these systems must respect people’s privacy.”
At the same time that OpenAI has begun touting its new AI features, an influential commentator has taken aim at why AI is so popular among some companies, especially when it comes to monetizing data.
“It requires the surveillance business model; It is an exacerbation of what we have seen since the late 90s and the development of surveillance advertising. “I think AI is a way to strengthen and expand the surveillance business model,” said Meredith Whittaker, president of messaging app Singal.
“The Venn diagram is a circle,” he added, speaking at TechCrunch Disrupt 2023.
“And the use of AI is also vigilant, right?” she said.
“You know, you walk past a facial recognition camera that’s equipped with pseudoscientific emotion recognition and produces data about you, right or wrong, that says, ‘You’re happy, you’re sad, you have a bad temper, ‘you’re a liar, you’re Ultimately, these are surveillance systems that are marketed to those who have power over us generally: our employers, governments, border control, etc., to make determinations and predictions that will shape our access to resources and opportunities”.
