Coca-Cola dropped a new batch of AI-produced ads, with a list of caveats a mile long. The beverage giant received pushback for its AI ads last year and took great pains to say that 100 people worked on the new AI ads, a number that they claim is on par with productions in past years, and that “a “team of artists” “work[ed] frame-by-frame, often pixel-by-pixel” to touch up and tweak the festive images generated by the AI.”
Pity the poor CEO trying to navigate the choppy waters of AI adoption. On one hand, big brands – like Walmart – are going in on it, but on the other hand, employees and customers are pushing back by “boycotting AI,” although this often proves to be functionally impossible in real life.
Recent rounds of layoffs at large companies and repeated pronouncements from AI leaders that even more jobs will be displaced have soured the mood of many; the feeling reignited in San Francisco when a cat was hit by a Waymo, and people started calling for the cars to be pulled from the streets.
All to beg the question – why use AI at all?
Indeed, the morass Coca-Cola stumbled into can be seen across industries and reflected in public opinion. The central tension seems to be: if AI is supposed to make everything better, why does it really feel like things are so much worse? Forget layoffs or propaganda – have you ever tried to use a bot for customer service? Tried to get ChatGPT to complete a task it swears it can do, and then hours later, reveals it really can’t (hot tip, it can’t edit your decks)? Asked it for restaurant recommendations only to discover the spot closed nine years ago?
When the news broke that 95% of GenAI pilots fail, some people celebrated this as the end of the AI era, but they are way off base, not least because AI companies are basically holding up the economy. Instead, there’s a simpler lesson here – companies are flailing and projects are flailing because they forgot one very important ingredient – solving problems.
Take the Coca-Cola example. Assuming they are telling the truth about the number of people they are employing to create the ads, they’re likely not saving much money. The ads are fine and quite cute, but don’t represent a marked increase in quality from previous generations of holiday commercials. They claim the production pipeline has sped up by using AI, which could generate savings – but is that scalable across an entire organization or a one-time, marginal cost decrease?
And it’s not just Coke tweaking at the margins. I was on a call with a huge company earlier this year who were effusively praising their amazing AI transformation efforts; it was exciting, except for the fact that they never specified what the efforts actually were. When someone finally asked, they explained that their corporate wiki was now powered by AI. While corporate wikis can be quite useful, it’s not exactly a flying car moment, and they also failed to explain why the old wiki was not working.
If companies want to succeed in using AI (or any emerging technology), they need to start by defining a problem that needs to be solved. Why weren’t the old ads working? Why didn’t our training lead to good outcomes? Why did the expense management software make people want to tear their hair out? They can then go deep and make real investments in building a solution, not just talking a big game and then quietly slinking away.
By failing to clearly outline how people’s lives will be improved by AI, companies risk more resistance and blowback when they do adopt it. Good AI-powered customer service agents could mean shorter wait times and faster answers for customers, as well as cost saving for a company – and the convenience would likely make any job losses more palatable to the public at large. AI job screening that effortlessly matched talent to teams could make everyone’s lives easier; AI job tools that sent resumes into black holes while leading frustrated recruiters to just give up and hire the boss’s nephew: not so much!
Companies need to come up with better stories and strategies around AI adoption, and quickly. It will benefit them as they discover real savings and learnings, and benefit the public as they see real benefits from automation. Otherwise, we’ll just be stuck in an endless loop of outrage and slop.
Read the full article here
