What are the pitfalls of AI product advertising pointed out by the FTC?



With the advent of ChatGPT for interactive AI and Stable Diffusion for image generation AI, the term ``artificial intelligence (AI)'' has begun to be used as if it refers to a versatile entity that can do anything. However, many of the products that actually incorporate AI do not produce the expected effects. Attorney Michael Atleson of the Federal Trade Commission (FTC) Advertising Practices gave an opinion on such a situation.

Keep your AI claims in check | Federal Trade Commission

https://www.ftc.gov/business-guidance/blog/2023/02/keep-your-ai-claims-check

Mr. Atreson started with the question, 'What exactly is AI?' AI usually refers to various technologies and tools that perform tasks such as prediction, judgment and reasoning on behalf of humans, but it is a vague term with many definitions. 'One thing is for sure, it's a marketing term. It's the hottest word right now,' Atlerson said. We are aware that the abuse is no longer put on hold,' he said.



The FTC has previously warned companies to avoid using automated tools that have a biased or discriminatory effect. However, the reality is that some of the products that claim AI are out of the question because they do not work as advertised in the first place. The FTC is also tasked with managing and controlling false or unsubstantiated claims about a product's efficacy, so advertising can move the FTC, Attleson said, noting that: I called.

◆Are you exaggerating what AI products can do?
Atleson points out that the FTC may take issue with exaggerating what products and services can do and claiming that they can exceed the capabilities of current AI and automation technologies. For example, product claims may not be scientifically backed, or claims may apply only to certain types of users or under certain conditions.

◆ Do you promise that your AI products are superior to non-AI products?
It is not uncommon for advertisers to say that some newer technology makes their product better, perhaps to justify a higher price, but such comparative claims require sufficient evidence. ``If you can't show evidence, don't make such claims,'' Atlerson said.

◆Are you aware of the risks?
If AI fails or yields biased results, we cannot blame the third party who developed the technology. Nor can companies be blameless just because they don't understand the technology, or because AI is a 'black box' and don't know how to test it.

◆Does the product actually use AI?
Just using AI tools in the development process does not mean that the product has AI. ``The FTC investigation allows us to analyze the materials and see if the content of the product matches the company's claims,'' Atleson warns to reconsider if he makes unsubstantiated claims. Did.

in Web Service, Posted by log1p_kr