As adults, we make thousands (and thousands) of decisions a day. However, if we don’t have the right information or data, it can be difficult to make decisions. The same could be said about artificial intelligence: The technology is requiring us to make decisions with sometimes limited or incorrect information or data, and it’s not working. We need to develop better datasets for AI to do its job better.

This solution sounds simple but for most advertisers and creatives, these datasets either don’t exist or worse, the same datasets are being recycled, leading to ineffective, inauthentic, and misleading results. In advertising, this can hurt or kill a campaign regardless of the brand or personality behind the idea.

Another thing that can hurt or kill a campaign is when brands aim for efficiency over effectiveness. While efficiency is important, effectiveness is a pivotal component for any ad regardless of whether you use AI or not. Brands are looking to companies like Google to optimize ads, but these platforms are only measuring and monetizing clicks, which is great for the platforms and terrible for brands. One example of this is Google’s latest announcement that it will rollout novel ads based on creative content. This only goes toward short-termism and is part of the problem.

In case you missed it, Google announced in April that it will be rolling out “novel” ads based on “creative” content. This content (images, videos, and texts) will be supplied by marketers and Google will then remix the material using AI to produce ads based on objectives like sales.

Google’s key performance indicators (KPIs) are to get clicks or interactions and these new novel ads will be altered to reach those KPIs, removing opportunities for brand building. This will hurt brands in the long run, especially when you consider what makes a great ad.

• Is it meaningful?
• Was it a well told story?
• Was it emotional?
• Was it different from other ads?
• Is it distinctive?

Google’s new approach will alter ads to drive sales and clicks. My question is: Why can’t KPIs include brand building? Why can’t we have both in ads?

These ads will have a very limited palette of creative data and will focus on things like the size of the logo and price point. This is limiting and prevents advertisers and brands from truly understanding effectiveness.

If we go back five years ago, most creative agencies were looking at datasets that focused on how consumers react to communications. This approach didn’t work then and it doesn’t work now. When a technology doesn’t exist, it’s incumbent upon leaders to create it themselves. For me, as a leader in the advertising industry, I want to create the right technology to deconstruct and understand the communications in a brand’s category—enabling the data to better inform the creative.

When you create specific AI tools built for advertising, as we did, you help drive the future of work for our industry. AI assistants will be essential for understanding the factors that inspire consumers to like or buy products, and the advertising industry overall will need to incorporate more creative data into the decision-making process.

We are still at that juncture when AI and advertising are at very different stages of regulation. The former is not regulated at all, and the latter is highly regulated with its own set of standards. If brands entrust AI to write an ad for you, you will undoubtedly encounter issues and problems. AI might be able to write some level of copy for ads, but it will also make things up; it will fill in blanks that aren’t true and that will get brands in trouble. In our industry, we must verify everything we say in an ad. AI won’t be able to do this.

It’s time that our creative leaders take control of brand storytelling and AI because we understand it better than anyone else. If we don’t, the platforms like Google will fill the void and claim authority on what makes a good campaign, and that’s not good for brands, agencies, or the real creatives that do the hard work of conceiving something new and different.