Should marketers be excited or concerned about ChatGPT?
It's important to understand ChatGPT is a ‘regurgitator’, not general intelligence, which is likely more than five years away, writes James Dixon, Atomic 212°'s chief digital officer. Let’s be clear from the start: this opinion piece is not written by ChatGPT, but it could well have been!
ChatGPT could well be a glimpse into a not-too-distant future where artificial intelligence (AI) can converse, inform and educate our human minds – even offer an immediate solution to my daughter’s Year 12 English essay.
The model is the creation of OpenAI, a San Francisco-based tech company that wants “to ensure that artificial general intelligence benefits all of humanity”. Investors include Silicon Valley luminaries such as Reid Hoffman, Elon Musk and Peter Thiel.
OpenAI’s biggest benefactor, however, has been Microsoft, which invested $1 billion in the company in 2019 and helped facilitate the ChatGPT project with training on Azure AI supercomputing infrastructure.
If you haven’t tried it, ChatGPT is free and accessible via a registration link here.
So, having seen 1 million users sign up in under a week – a number of users, for reference’s sake, it took Facebook ten months to achieve – is the hype merited?
In its favour, ChatGPT communications give the impression of an intelligence on par with our own, providing responses that are original content – no copyright pertains – thus allowing the user to put this infinite copy into action for a wide variety of purposes.
It is able to generate answers to questions, create fictional prose, even write code.
This opens new possibilities for marketers in terms of time saving (copy writing, customer research, website and app builds), creative and product design (generative AI visuals), as well as economic and scalable customer service (intelligent chat and call centres).
But can this intelligence replace the marketing function in part or whole? Are our jobs and skills at risk?
You can relax for now – the expert view is that we are some five to ten years off AI becoming ‘general intelligence’.
To illustrate this further, the analogy is that the internet is the library of human knowledge, Google is the librarian, and ChatGPT is a regurgitator of the information in the library – it can only serve from the past bank of information.
And by past, I mean prior to our current decade, with the program’s FAQs stating “It has limited knowledge of world and events after 2021”.
Take, for example, when I asked ChatGPT the question: “What products are trending now?”
ChatGPT’s response began with the clarifier, “I am unable to provide specific information about what products customers want at the current time, as I am a large language model trained by OpenAI and do not have the ability to browse the internet or gather information about current market trends.”
I’d say that’s important info for the user to be aware of in most interactions – although fortunately, according to OpenAI, this kind of admission comes up frequently: “The model is often excessively verbose and overuses certain phrases, such as restating that it’s a language model trained by OpenAI”.
Following its disclaimer about being unable to provide real-time updates on consumer behaviour, ChatGPT attempted to tackle my question about trending products: “The demand for specific products can vary widely depending on a range of factors, including the current state of the economy, changes in consumer behaviour, and the availability of new and innovative products.”
It’s not wrong, but it doesn’t answer my core question. It doesn’t nail the brief.
The bigger concern is when it actually is wrong – which it frequently is, OpenAI noting “ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers.”
So as useful as the ability to create cohesive and (reasonably) informed responses communicated in eloquent copy may be, ChatGPT can’t be relied upon to replace… well, anyone just yet.
Its ability to access current information and provide more accurate responses will improve over time and, in fact, helping speed this process along is part of the reason it is being offered for free at the moment.
OpenAI want as many people as possible to use the model to help improve it, although CEO Sam Altman made it clear on Twitter that the company will have to monetise the model at some point to offset its “eye-watering” costs.
However, even after the issues of timeliness and accuracy are resolved, ChatGPT will still struggle to build more than a passing affinity with humans.
It’s hard to get angry, happy or emotionally aligned to a chatbot, which therefore means it’s hard to imagine brands can be built by these interactions.
ChatGPT is an amazing tool that can summarise historical information into neat, sometimes witty dialogue, but ultimately it can’t relate to its subject with empathy, imagination, tact or diplomacy.
It is a servant, not a master.
General intelligence AI is the moonshot ambition, whereby the code becomes sentient, able to feel its responses and adjust them to the human interaction.
The jury is out on whether this is possible, with most experts saying that several iterations of computing mechanics are needed to deliver it. We are not currently at that I, Robot moment.
At the core of marketing is the understanding of human behaviour; emotion; desire. These are non-binary traits that ChatGPT cannot ‘solve’.
So while the media coverage has been prolific and sensationalised, AI is still a way away from being able to inspire the kind of emotion that brands require to be successful.
All that said, this is not a technology to be dismissed simply because it’s not perfect. It may not be ChatGPT that we are talking about in five years – perhaps remembered instead as the Myspace of social media that paved the way for the likes of Facebook – but generative AI will have a role in our industry.
That role will be prominent. And it will be soon.
James Dixon, Atomic 212°’s chief digital officer
I think the most interesting point is ‘why’ ChatGPT can’t interface with more recent events.
It looks like it’s because the model is highly trained and calibrated. It’s a model’s ability to observe new information and calibrate to it that makes it unique, not a model’s ability to exist at a fixed point in time.
Frankly ChatGPT is an impressively large, but highly fragile, model. Super cool for lots of things. But largely an interface to stuff that’s been around for a while than a great leap forward in AI.
User ID not verified.