.gif)
OpenAI's latest ChatGPT update bestows the AI chatbot with the gifts of speech and sight.
What happened: ChatGPT can now converse with users — across five chipper voice options — and process images to help answer questions. The features will be available to paid users within two weeks and to those of us who are AI freeloaders in the near future.
- The chatbot can leverage personal information and, per The Wall Street Journal, boasts voice capabilities that are “almost indistinguishable from a human at times.”
Zoom out: In addition to giving ChatGPT a voice, OpenAI is teaming up with Spotify to clone the voices of podcast hosts to generate foreign-language versions of their shows, with some of Spotify’s most popular podcasters, including Lex Fridman and Bill Simmons, signing on.
Why it matters: As ChatGPT adds new functions, it can gather more info and become more well-rounded. As it does this, it’s increasingly starting to look like a wide-ranging consumer product that will compete with top voice assistants like Google Home and Alexa, per WIRED.
Competitors are on the offence.
- Google has been testing new ways to utilize its AI assistant, including experimenting with creating a robot life coach.
- Amazon recently previewed a language model that promises to give Alexa “near-human-like” conversation skills.
Yes, but: With every new development comes a new set of risks. ChatGPT’s newfound ability to synthesize voices comes with pretty obvious use cases for abuse. Not to mention the age-old concerns about data privacy vis a vis giving a robot a bunch of personal info.
Bottom line: It has somehow only been ten months since ChatGPT debuted to the general public. If the rapid advancements in that short timespan are any indication, we’re not done seeing wildly impressive/kinda scary innovations from OpenAI any time soon.—QH