The launch of a new “empathic voice interface” by Hume AI is certainly a bold move in the world of artificial intelligence. Promising to add a range of emotionally expressive voices and an emotionally attuned ear to large language models from well-known companies like Google and OpenAI, this technology signals a potential shift towards AI systems that can better connect with users on an emotional level. However, a critical analysis of this development is necessary to fully understand its implications and limitations.
The claim by Hume AI co-founder Alan Cowen that their technology specializes in building empathic personalities that speak like people rather than stereotypical AI assistants is dubious at best. While it is true that their voice technology, EVI 2, has been found to be emotionally expressive, comparing it to OpenAI’s ChatGPT, which has faced its own controversies regarding voice imitation, does not necessarily validate Hume’s claims. The reference to a “flirtatious voice” used by OpenAI and the subsequent comparison to Scarlett Johansson’s voice raises concerns about the ethics and originality of such technologies.
While Hume’s voice interface may be more emotionally expressive than traditional voice interfaces, the technology’s ability to accurately measure and respond to user emotions is questionable. The mention of values like “determination,” “anxiety,” and “happiness” being displayed during interactions raises concerns about the accuracy and reliability of these measurements. Additionally, the ability to simulate specific emotions like being “sexy and flirtatious,” “sad and morose,” or “angry and rude” may come across as gimmicky and superficial rather than genuinely empathic.
The demonstration of Hume AI’s technology to students by Professor Albert Salah raises hopes for the future of affective computing. By assigning emotional values to users and modulating speech accordingly, Hume’s technology has the potential to create more human-like voice interfaces. However, the occasional glitches and odd behaviors observed during testing, such as sudden speeding up and gibberish output, indicate that the technology still has a long way to go in terms of refinement and reliability.
While the concept of an empathic voice interface is intriguing and has the potential to revolutionize the way we interact with AI systems, the current limitations and questionable claims surrounding Hume AI’s technology raise significant concerns. As technology continues to advance in the realm of affective computing, it is essential to approach these developments with a critical eye and a focus on ethical and user-centered design principles. Only then can we truly harness the power of AI to create more empathic and human-like interactions in the digital world.