AI Mentions May Not Translate To Trust, New Analysis Suggests

Brands tin look successful AI-generated answers and still not travel crossed arsenic believable, according to an analysis from communications agency Burson.

The report, called “The Credibility Paradox,” says getting mentioned successful AI answers isn’t the full picture. Burson argues that brands should attraction astir really convincing those answers really are.

For transparency, Burson sells estimation consulting and GEO services. The study was produced pinch Profound, an AI trading platform, and uses Decipher, a exemplary Burson built to foretell really convincing each reply would be.

What The Report Found

Concrete claims tended to beryllium rated higher than absurd ones. When AI platforms responded to questions astir a company’s products aliases workplace culture, those answers often felt much believable compared to responses astir governance aliases leadership.

Business group were much easy convinced. The exemplary rated AI responses 10% much reliable for business audiences than for others. Business audiences cared astir about innovation, while consumers were astir willing successful a company’s workplace civilization and products.

How The Analysis Worked

Burson asked 7 AI platforms questions astir 85 companies, past ran the answers done Decipher, a exemplary that predicts really believable each consequence would be. That produced much than 55,000 scores.

The scores are predictions, not responses from group who publication the answers. The study doesn’t people its prompts, institution list, aliases really Decipher’s scoring works.

Why This Matters

If you’re search your brand’s beingness successful AI answers, you’re measuring the easy part. This study suggests that what the AI says astir you mightiness matter much than the mention itself.

Google precocious called GEO “still SEO.” This study makes the lawsuit that there’s a furniture group aren’t looking astatine yet, and that’s if the reply is convincing.

Looking Ahead

The predictions haven’t been tested against really group really react, which limits what the findings tin show us.

The findings propose that paying attraction to what AI says astir you, not conscionable that it mentions you, is worthy watching arsenic it plays a bigger domiciled successful really companies get discovered.

Featured Image: elenabsl/Shutterstock