The Robots Are Here!
Have you heard about this thing called ChatGPT?
Just kidding. Of course you have.
Unless you’ve been living under a rock, snoozing Rip Van Winkle style or Doomsday prepping on the dark side of Mars, there is absolutely zero chance that you haven’t encountered at least 2,578 media mentions of this super secret new technology that only a handful of humans (as in 7.888 billion) have been made privy.
That said, it’s kind of a viral topic.
So what else is there to say that hasn’t been said thousands of times by humans and the AI itself.
Probably not much. But I’ll give it a shot.
First however, a very brief primer on ChatGPT (because you likely know this already).
What it is:
ChatGPT is an artificial chatbot. You can ask it any number of things and expect a conversational-like response. Unlike traditional Google, Yahoo! or Bing searches that only serve up links to content, it creates the content itself using the power of machine learning. For instance, I asked this rhapsodizing robot to explain itself. Et voilà:
“Hello! I am ChatGPT, a large language model developed by OpenAI. I am designed to understand and respond to natural language inputs in a conversational way. I have been trained on a vast amount of data and can provide information, answer questions, make suggestions, and even engage in casual conversation. My purpose is to assist users in various tasks, such as helping with research, providing recommendations, or simply providing entertainment.”
What it isn’t:
Human. At least not yet.
What it all means:
John Oliver said it best in his excellent episode on this topic: “If AI can do what humans can do but only better, and I’m a human, then what happens to me?”
The end of the line for many white collar jobs. And the beginning of many more. At least that’s how most disruptive technologies tend to play out. Yesterday’s most innovative buggy maker is today’s Elon Musk. Or something like that.
Anyway, I’m no futurist. But I’m very interested in where things are headed, particularly on the digital marketing front. So let’s get to it…
Right off the bat I have some very good news!
I broke ChatGPT with a simple question!
Crisis averted. Disrupting disrupted.
All kidding aside. The rise of AI, machine learning and ChatGPT-like bots is a really big deal. But how big?
If you follow Christopher Penn (co-founder and chief data scientist at TrustInsights.ai), you know that ChatGPT is already having a significant impact, particularly regarding content development.
That’s primarily because the tool is good at editing and modifying something that’s already been created. ChatGPT’s ability to refine messaging is a huge advantage.
It’s also incredibly quick. Digital marketers can produce content in minutes using ChatGPT, instead of spending hours or even days crafting and refining messaging. This is especially valuable for social media postings, which often need to be created and shared quickly for success.
This ability to churn out massive amounts of content is perhaps the biggest eye opener. While mass quantities of content have already become commoditized, AI is poised to change the nature of its development. This will have far reaching effects beyond social media strategies. For instance, how will search engines handle this inevitable content infusion? Will Google penalize AI-generated messaging that’s clearly designed to boost page rankings? It wouldn’t be too tough for a dentist or a plumber or a lawyer to use tools like ChatGPT to write scores of articles on their specific services or specialties and then discretely post them in their respective website news sections, possibly back-dated to keep this content “discoverable” by crawling robot spiders but mostly hidden from human visitors.
Google and its search competitors will no doubt fight fire with fire, using AI to combat AI. After all, ChatGPT detection tools hit the market almost as quickly as ChatGPT became a household name. OpenAI, the creator of ChatGPT has even released its own "OpenAI Text Classifier” to help distinguish between AI-generated and human-generated text.
It’s also not hard to imagine ChatGPT being integrated into Google Ads. Marketers would simply input keywords and preset guidelines and AI would do the rest—create titles and descriptions, and improve the ads that aren’t performing well. This is already happening on the back end, with Google’s algorithms selecting amongst human-produced headline and description options to determine the messaging that best resonates with searchers. Soon enough, AI will be writing for AI.
In fact, the integration of ChatGPT into other technologies—inevitable from the start—is already underway. OpenAI has just introduced an API (essentially a programming interface allowing two or more computer programs to communicate) that lets businesses build ChatGPT tech into their apps, websites and products. Early adopters include Snap, Instacart and Shopify. Of these, Shopify has moved the quickest, already integrating ChatGPT into its AI-powered personal shopper, enabling its 100 million or so users to chat about everything from gift recommendations to home decor ideas.
Yet as exciting as things are on today’s AI front, here’s the rub. It’s still early in the game and awkward bugs abound. Since ChatGPT has been trained on data that includes large swaths of internet content, its output can sometimes reflect that. One inevitable result? Some users have experienced racist and sexist answers to questions they’ve posed. ChatGPT can also invent facts—a fact that has been openly acknowledged by OpenAI's chief technology officer Mira Murati.
And therein lies the Achilles heel of ChatGPT and its brethren. They’re only as good as their training. When exposed to sources like Reddit discussions, all bets are off. And while ChatGPT’s content output can be quite impressive, its prose is not (at least not yet). Many find its rules-based, standardized form of writing formulaic in structure and style. In other words, robotic.
So where does that leave us?
Here’s what I think:
Ultimately, human creativity can’t be replaced. But it can be enhanced. And that’s what tools like ChatGPT are already good at (and they’ll only get better).
Just to be safe, I asked the AI oracle itself for confirmation:
See. One can always trust a robot.