Estimated read time: 7m 20s
With over 7 years of cross-channel experience in digital marketing, Matthew Howman, Marketing Strategy Director at Affinity regularly contributes to the blog on a diverse range of topics, from analytics to hands-on marketing advice.
Over the past few months, you will likely have heard a reference to a certain Chat-GPT. It’s the first piece of technology that has brought AI chatbots into the forefront of the media, and the buzz that’s surrounded it has propelled artificial intelligence use cases to the top of everyone’s agenda.
However, Artificial Intelligence isn’t a new concept. Back in 1997, IBM launched Deep Blue, a supercomputer that was able to defeat the then World Chess Champion, Gary Kasparov.
Fast forward to 2023, the acceleration in capabilities has been primarily due to increased computing and processing power, a significantly lower cost of accessing that computing power and the layering of machine learning on top of artificial intelligence that’s made it an immeasurably more powerful beast.
AI is no longer simply a rule-based system designed to find the perfect outcome, but the machine learning element gives it the ability to absorb millions of data points and improve outcomes over time.
The tool that everyone is currently talking about, Chat-GPT, is a language processing tool that essentially brings words together and uses data to predict the best outcome in the sequence, producing some pretty outstanding results.
So what have we learned so far? Well, if you follow developments in the news and on Twitter, you’ll see that people have put AI tools to good use; fake image generation of Donald Trump being arrested, songs written in the style of Nick Cave and most importantly using it to get away without having to do any homework.
People are excited about artificial intelligence, Chat-GPT, and other large language models because of the near-unlimited possibilities they offer for changing everyday life. If you take the homework story as just one example, the addition of AI in day-to-day life and education is a potential game-changer.
With chatbots able to process complex queries and output very considered answers, you could have a scenario in the near future where every person has their very own personal teacher/assistant, able to gauge their current level of understanding on a topic and provide step-by-step guidance to help them improve their knowledge to a certain standard (e.g. GCSE, A Level, Degree etc).
I saw an example the other day where the chatbot refused to answer a mathematical problem, sticking within the parameters of helping the individual think objectively about how to solve the equation, rather than simply giving them the answer – a massive step forward in comparison to a query and result based system that we currently get from using search engines.
It’s also worth reading Bill Gates’ piece The Age of AI Has Begun, where one of the more interesting points he considers is how artificial intelligence could be used to solve some of the world’s biggest health issues, noting the death rate statistics of under 5s in poor countries.
How do you make decisions that can affect millions of people? By being able to cut through significant quantities of data and learn over time to improve the results.
There are still teething problems and a long way to go in the journey of AI taking over our day-to-day lives. In the case of GPT, the results aren’t always perfect, the data is limited to a certain timeframe and mistakes have been pointed out.
In the wider scope of artificial intelligence, bias can still creep into answers, which raises philosophical debate around political and cultural influence. In short, we’re still on the cusp of the revolution.
Artificial Intelligence & Machine Learning in Marketing
Cutting to the world’s most important use case for artificial intelligence – online marketing. AI has been in play for several years, and we’ve seen many influences develop across our business during this time.
One of the biggest changes I’ve had to adapt to is machine learning across paid search advertising. If we look specifically at search engine advertising, how we buy media has gravitated away from a traditional auction-based bidding system.
Instead, a large portion of media buying is now done by employing machine-learning bidding strategies that are designed to learn over time to reach certain goals.
As an example, to advertise on a certain keyword e.g. “wooden sheds” you might now choose to set machine-learning on its way to achieve a certain return on ad spend.
So rather than choosing to bid a certain monetary amount within the auction, it can be flexible while the algorithm finds the sweet spot that will achieve a certain level of return, as opposed to being focused on a fixed cost of a click.
So – what’s the future?
With the rollout of Bing Chat and Bard (Google), the most interesting change I can foresee is not in the processes in which we as marketers work, but in how people engage with businesses and shop on the internet.
If you consider the last 10 or so years, it’s been the era of the search engine. Type in a question or search for a product or service using a keyword or phrase and a search engine like Google will churn out a page of results. There are a lot of nuances to that depending on the type of query, but broadly speaking, that’s how we’ve found information online.
With AI or specifically an AI chatbot, there’s a level of engagement that search engines can’t replicate. If I use my wooden sheds from earlier as an example…
Typically a user might visit Google and search “wooden sheds” and then depending on the results, either visit a site or perhaps re-search with some more detail e.g. “pressure-treated wooden sheds” until they find a product or manufacturer that takes their interest.
With a chatbot, the journey will become more conversational and, with the opportunity for added complexity becoming familiar and part of the expectation, user queries will develop over time. For example, you could ask it to check reviews, pre-qualify pricing, or layer in the requirement for free delivery.
The above example was the first result I received from Bing Chat – the follow-up question will narrow the search further but rather than focusing on ranking for a particular query, you might find that in this method of searching, eCommerce marketers will need to focus on optimising for features and benefits as well as the product itself.
Over time I can see things developing into full conversations – again, a personal assistant. A few years ago everyone made a lot of noise and hype around voice search becoming the next big thing.
It would never take off because a) it only really works for question-based searches rather than commerce and b) no one wants to sit in public having a conversation out loud on their phone, asking it to find products to buy.
The key difference between a chatbot and anything that’s come before it is that it can become more intelligent over time. It will be exposed to billions of searches, questions, and conversations every day – and the more data it’s exposed to, the more intelligent it will become.
It’s also perfectly suited for the generations that have grown up alongside the internet, search engines, social media and the iPhone.
What does it mean right here, right now?
In all honesty, not a lot. Because Chat-GPT hit mainstream news, there’s been an initial frenzy around its potential, but even the creator of Open AI has indicated we’re way off finding its true potential.
One indicator of the herd mentality was Buzzfeed announcing they would interlace Chat-GPT with their content generation, the stock price flew up from less than $1 a share to almost $4 and within a week it came tumbling back down again.
There’s also been a lot of noise made by big names in the tech world calling for regulation and a halt to AI progression, due to the potential threat to society.
What we might find is a more staggered approach to the developments in chatbots over the next few years as the usual government bureaucracy and need for control weaves its way into the ecosystem.
However, keep an eye on how things progress – these are early days, and while it might take a few years to kick in, the marketing industry will change substantially as it does.