How I Learned to Stop Worrying and Like AI

As a writer, marketer, and storyteller, the latest generative artificial intelligence and its surrounding furor feel personal. Are these tools destined to outpace the skills I’ve spent decades acquiring? Will I become no more than a typist as networked programs surpass the capacity of my human brain? When the machines are the creatives, do we do the work of the machines?

This little existential crisis could have turned into an all-out spiral if, in my role as a journalist, I didn’t have access to thinkers who are much more grounded in the day-to-day realities of AI. 

One such realist is John Cosley, Senior Director, Brand Marketing, at Microsoft Advertising. Microsoft has been at the epicenter of the conversation on AI because of their investment in OpenAI, the company behind ChatGPT and DALL-E, two of the most discussed and advanced programs in the space. 

"Microsoft has been pretty intentional in its position that we believe AI is going to usher in a new era of productivity for businesses and for the economy as a whole,” says Cosley. “At Microsoft we’re not thinking, ‘Hey, let's just do sparkly fun innovation and just be disruptive.’ We're trying to think about how we bring viable, commercially ready innovation to market, so that way we support an ecosystem.”

Copilot is a great example of this AI ecosystem that Microsoft is creating. Microsoft 365 Copilot is billed as an AI-powered assistant which utilizes large language models (LLMs). It's a tool that Cosley himself employs. 

“As a communicator and a brand marketer, words are important to me, and thinking about how to say things as clearly and specifically as possible. It's great to have Copilot there to help find ways I can think about something differently,” says Cosley. He also uses Bing Image Creator to kickstart his creative process and LLMs to to expand keyword lists, and synthesize and summarize reports.

Cosely is sanguine about AI in part because in his work at Microsoft Advertising, AI applications–including some generative AI and large language models–have been built into the ad tools for a while. 

“We have been building AI into our ads platform for a long, long time,” Cosley says. “It's in our audience intelligence and how we take data signals and create audience segments like in-market audiences. It sits in our marketplace in terms of how we do matching between ads and queries. It's in the tools and capabilities that are available today, such as responsive search ads and dynamic search ads, and how you can auto create content.” 

Less flashy than the customer-facing products like ChatGPT and DALL-E, these tools may not attract the media attention but are equally transformative. And the way in which they are woven into existing workflows and products may be more indicative of how AI will appear almost imperceptibly around us.

This indistinguishability is what gives Alex Kauffmann, former technical project lead at Google and current AI consultant for projects like the 2022 sentient killer robot movie M3GAN, pause.  

“It becomes much easier and much cheaper to generate marketing, marketing content, copy or images or whatever,” Kauffmann says. “There's a sense that maybe when the content creation becomes so cheap, nobody's actually going to pay attention to it.” 

The ubiquity of easily generated content will force us to create new formats, Kauffmann thinks. Image and text will be so readily available, so customizable, that to attract customer attention, marketers will need to create new content buckets. 

“You can make ads that are really, really targeted. And that can be a little bit creepy,” Kauffmannn says. “You can think about customization in a way that you've never thought about it before. When you have a more nuanced understanding, you can take advantage of actually being able to look at what customer info you have and generate something on the fly, rather than using a set of already created assets that you're combining.”

Kauffmann sees this extreme customizability seeping into all mediums, into movies where product placement will be served based on viewer interest, as well influencer marketing where Kauffmann imagines the influencers themselves being creations of AI. 

“Influencer marketing is, in a sense, a way of overcoming a trust problem,” Kauffmann says. “On Instagram, you have people that you follow and you know about them and they say, ‘Hey, try this thing.’ I believe that now what happens with all this generative stuff is that you can generate an influencer. I can make a video of a person who looks like a totally real person saying something totally believable that's completely generated. And at that point, that token of trust is gone.” 

While he shares his concern about dissipating trust and AI-generated content farms, Kauffmann easily puts to rest any fears of a computer takeover or retaliation. 

“My job on M3GAN was mostly to remind the writers that it should be a robot. They're like, ‘Oh, she's so upset,’ She’s not upset. She's a robot,” he says smiling. “She simulates feeling maybe, if you want, but that's not the same thing. She has no internal motivation.”

And this is ultimately the thing that makes me feel better about my job and our future of AI-assisted marketing: artificial intelligence can take orders, it can generate the best answer, maybe even an exceptional one, but it has no inner emotional life or motivation. Everything that it does comes from us. We will have to keep feeding the machines, our thoughts, our schedules, our queries. 

“We're still learning how consumer behavior is evolving and how consumers engage with chat versus traditional search,” Cosely tells me to explain how consumers are reacting to Microsoft's new programs. He explains that because of the multi-question search functionality how people ask questions in chat can be different than how they do it on just a general search, reflecting the reality that AI is a contextual mirror of our desires, who we are and what we want. 

“Everybody hedges that there's a 10% chance that it’ll go off with the paperclips,” Kauffmann says, referring to the fabled Paperclip Theory scenario, where the robots eliminate the humans, who they perceive as an obstacle to their mandate of creating the most paperclips. “I think there's no chance. I think the future is always much more banal and much, much different than what we imagined.”

‍

Magnet is a global brand studio that uses data and storytelling to drive measurable business results. Our team is made up of strategists and creatives who use our THINK / MAKE / REACH process to develop marketing strategies, world-class creative production, and distribution strategies for our clients. Our work starts with data and insights to develop storytelling that’s distributed across all platforms. Our clients are currently appearing in the top media outlets, being shared by influential community members, and rising to the top of the search rankings. Recently we've produced award-winning campaigns for Google, Adobe, Chase, Microsoft, LinkedIn, CHIEF, Greenhouse, Cedar, YouTube, PBS, HarperCollins, Goldman Sachs, Airbnb, IBM, and more.