Scarlett Johansson’s complaint to OpenAI is a new benchmark in the development of machine intelligence

From The Conversation: 2024-05-23 07:26:58

Over 2,000 years ago, Aristotle introduced the concept of rhetoric, using logic, audience understanding, and speaker authority for persuasion. The recent launch of GTP-4o by OpenAI utilizes these strategies, sparking excitement and concerns. Actress Scarlett Johansson expressed shock and anger at how the AI sounded like her in a specific role.

GTP-4o, different from traditional LLMs, can understand text, vision, and sound, recognizing nuances like emotion and integrating them naturally. Its unnervingly quick and natural responses, including singing, raise concerns about potential weaponization as government organizations grapple with its implications in combating deepfake threats.

Language models like GTP-4o present a new way to interact with AI, marking a shift in technology interaction known as Empathetic AI. As the speed of AI advancement grows, concerns arise about potential misuse in elections and fraudulent activities involving deepfake audio. The importance of not underestimating the risks of weaponized AI becomes increasingly evident.

Rhetoric and persuasive power have always been about not just what is said but how it is communicated. Unconscious bias about accents, as highlighted in a UK report, plays a role in perception and trust. AI in call centers is now being used to tailor voices, illustrating the importance of delivery in effective communication, as identified by Aristotle centuries ago.



Read more at The Conversation: Scarlett Johansson’s complaint to OpenAI is a new benchmark in the development of machine intelligence