Dove is a brand I’ve always admired - it has always been on the side of women, and actively explores how it can uncover our insecurities for what they are. That is to say, it puts social expectations and our expectations of ourselves under the spotlight and shows them to be undermining our worth.

And once again Dove has shown that it values real women by promising not to use AI-generated women in its ads. Dove estimates that by 2025, 90% of the content we see online will be AI-generated. Already, fake women abound on social media, and the results lean from the sad (they make us feel bad about ourselves) to the downright terrifying (deep-fakes, bots and more come to mind).

There are of course some really exciting ways to use AI, and even AI people - even if that does make me feel a little weirded out. But for any brand that stands on authenticity, and representing real people, it’s a hard line to tread.

Ultimately, the right approach to AI generated people will differ for each brand. But there are some universal principles:

Show and tell

Brands that wish to have a human face that connects with their customers should probably do just that - have real people talk about their business. But there are times when an AI avatar will do.

Synthesia has revealed AI avatars trained to display human expressiveness. This might sound like the start of what led to Blade Runner, but the company is deploying this advanced new tech responsibly. Only large (enterprise) businesses can access the tech, and human moderators check the content to ensure it’s not being used for misinformation. By and large, it appears that these avatars are being used for training videos internally.

But whatever the ways AI avatars are used, people need to know about it. Watermarks, declarations - however it’s done, it needs to be obvious. Obfuscation in this space will rebound on brands and possibly cause a headache far more expensive than the savings made in not paying an actor.

Check your bias

I don’t mean yours personally - though do that, too. Data owned by any business, or bought via consortiums or similar arrangements, builds in existing bias. And AI can compound and amplify that.

Crisis preparation means laying the steps that both avoid crises from occurring in the first place and by creating cultures able to deal with them. A business able to interrogate the biases presented by its data is taking steps towards both by encouraging open minds and mitigating future disasters.

This also means not overcompensating and landing where Google Gemini did, creating unrealistic and even offensive scenarios by deliberately diversifying images, without sensible thought into where and when that doesn’t work. Using AI images, AI content is a powerful tool, and like anything powerful can go really wrong. Making sure infrastructure is set up and biases are interrogated needs to be the foundation block of any AI based activities.

Value and prioritise human input

Whether an AI chatbot, an AI avatar or an AI background software; only human oversight can keep brands from blundering into reputational or operational issues.

Readers of Isaac Asimov’s I, Robot know that intelligent machines can have intentions of their own. We may not be there yet, but it’s already true that an AI might make decisions a human would not. Hiring talented people and trusting them as AI guides is essential for businesses that don’t wish to wander into an AI-generated crisis.

I still believe that humans connect best with humans, and applaud Dove’s commitment to this principle. But AI has a useful role to play, too - as long as brands always choose authenticity and stringent governance, and look for problems before they emerge.