ChatGPT has been capturing everyone’s attention on both sides of the pond, and the Crossmedia offices have been no exception. Here, Matt Asman, Director Analytics Operations at Crossmedia New York, shares his thoughts where we’ll see ChatGPT pop up in the immediate future and how marketers can use the tool to their advantage.
Listen to this article:
It’s been a little over a month since OpenAI’s ChatGPT launched. Aside from chatter of a $29 billion valuation, we’ve also seen a wealth of funny limericks, endless lines of code, and even entire blog posts being generated by what is effectively the most advanced chatbot ever made public. Is this the next big push in technological advancement? Will it change our ways of working and living forever? Are we at the end of human civilization, ushering in the robot singularity? And most importantly…how will it impact marketers? (Won’t somebody think of the marketers?)
A bit of background: GPT (which stands for Generative Pre-training Transformer) is a language model developed by OpenAI, an organization dedicated to researching AI projects involving machine-learning and natural language processing. While the model has undergone many iterations over the past few years, OpenAI launched a free demo of its latest and greatest chatbot at the end of November 2022 – dubbed ChatGPT – for anyone to use. It leverages one of the largest datasets of text to ever be ingested for this purpose, and the output creates eerily human-like responses to pretty much any prompt you can imagine.
The demo immediately went viral, and the results are indeed impressive. Here’s what it gives me when I ask it to write a Seinfeld scene where Kramer pitches Multi-Touch Attribution to Jerry and George:
(Jerry and George are sitting in Monk’s Cafe)
Jerry: So, how’s that ad campaign going for your new app?
George: It’s going alright, but we’re still trying to figure out the best way to measure its effectiveness.
Kramer (entering the cafe): Hey Jerry, George, what’s going on?
Jerry: We’re talking about George’s ad campaign.
Kramer: Oh, I’ve got just the thing for you. Have you heard of multi-touch attribution?
Jerry: Yeah, it’s like a way to measure the effectiveness of an ad campaign by tracking every touchpoint a customer has with the brand.
Kramer: That’s right. And with multi-touch attribution, you can see which channels are driving the most conversions and adjust your strategy accordingly.
Okay, so maybe it won’t crack your top 10 funniest Seinfeld episodes of all time…but we must hand it to the AI for accurately summarizing what MTA is and nailing some of the show’s details. Sitcom spec scripts aside, what do we see the GPT technology being used for in the immediate future? Here’s a few use cases within the advertising, marketing and analytics space that come to mind:
- The most obvious – support chatbots: Those annoying little chat pop-ups in the corner of your health insurance company’s website might be getting a whole lot more user-friendly. Rather than following the rigid logic of scripted or rule-based chats that regurgitate an FAQ, expect much more human conversations with your friendly neighborhood support AI. With this improved experience could come higher usage rates, which could spur a sizeable volume increase in these types of touchpoints on a brand’s site – capturing conversations that would have otherwise taken place offline. We may even see automated ways for language models to distinguish between qualified and unqualified leads, or to determine where in the funnel a potential buyer might be, helping to further personalize a customer’s experience.
- Light copywriting for search, display, email, and SEO campaigns – emphasis on “light”: Even though Ryan Reynolds tried to make it write a Mint Mobile commercial, let’s be real – the AI isn’t winning a Clio award any time soon. But for spitting out hundreds of search ads, email subject lines, or banner headlines, the tech immediately creates a solid block of ice for a professional copywriter to carve from – especially for those multi-variate testing projects where the more concepts, the better.
- Writing code – or rather, solving coding challenges quickly: ChatGPT knows almost every coding language from JavaScript to Python. Many analysts on Crossmedia’s Redbox team took to it as their own personal coding assistant and began solving problems and developing applications for creating efficiencies almost immediately. One Redbox analyst figured out how to create a Python application that merged files together for a faster ETL process. Another team member finished coding an automation project they had been struggling with for weeks. The efficiency and time-saving possibilities now seem endless!
It’s important to note that much of this technology (AI-powered SEO, for instance) has been steadily creeping to market since at least the past year. Machine learning and chatbots are nothing new, and this isn’t as much of a story about “emerging” AI as it is about access to a level of processing power that was once unfathomable.
There’s also many kinks and ethics concerns to be worked out too. The model certainly tends to spit out a lot of disinformation. Copyright concerns are already being discussed. And the programming Q&A site StackOverflow is even banning code that is seemingly generated by ChatGPT, because it is sometimes flat-out inaccurate.
At the end of the day, the GPT is best seen as a tool – like a calculator – that can assist marketers and help them view challenges from a new perspective, rather than a complete replacement for human brain power. While AI can crank out punchy ad copy and help with SQL queries, we will always need human thinking and collaboration to create compelling narratives, insights and strategies. It’s up to you to play around with ChatGPT to see where it fits into your own specific toolbox.
If you’re curious about how our German colleague Dirk Nögel feels about ChatGPT, you can find his musing here.
What Do You Think?