With all the buzz around AI and machine learning, it can seem impossible for many marketers to separate the signal from the noise. Is it the best thing since the internet? Are we all out of a job? As with most new technologies, the truth is slightly less dramatic (at least in the short term), and it helps to be armed with some basic knowledge to cut through the noise.
In reality, these technologies are more and more part of our everyday lives and, most vividly apparent where brands connect with consumers, from Amazon’s Alexa to online chat support. These are just two customer touch points that marketers should be aware of the impact of AI. There are many more, ranging from how and when your ads show up online, to revealing new customer segments that you might market to, or even develop new products for.
So, as we increasingly see this new technology insert itself into new parts of our lives, these are my top truths and myths to help marketers identify what’s worth paying attention to, and what’s just hype.
Three truths about AI
1. Anything that can be automated will be (and probably should be)
Marketers are drowning in data. To be effective, modern marketing needs billions of decisions to be made quickly and accurately every day. These decisions help determine which audiences see which messages and the right level of a client’s budget to assign. Humans cannot be expected to handle all of these decisions and thus we turn to machines. Machines excel at processing vast amounts of data and can analyse complex patterns at scale in milliseconds. What’s more, they’re excellent at identifying patterns that humans might miss, which helps identify emerging areas of demand that a brand may have previously overlooked.
2. When you say AI you probably mean machine learning
While both terms are synonymous for most people, knowing the difference will earn you serious points with any engineers you might work with. AI is a poorly defined term, referring to a field of research concerned with the creation of human-like intelligence. Machine learning is concerned with getting machines to complete tasks without being explicitly programmed to handle those tasks. These machines thus “learn” and get better at subsequent tasks with experience. An example of this would be Amazon’s Alexa. Alexa is not explicitly programmed to recognize all speech, rather the machine learning algorithms that it employs get better at understanding commands by improving its own recognition of different accents and common combinations through exposure to more and more examples.
3. Brands that make better use of first-party data will win
Most of us are familiar with the phrase ‘garbage in, garbage out’. The same holds true for machine learning, but more so. If we, as humans, are capable of making disastrously poor decisions based on incomplete data, imagine that scaled up to the speed and power of modern computing technology.
The biggest culprit of bad machine-led decisions - and cause of executive disappointment - is the use of a hodgepodge of second or third-hand data bought or borrowed from various partners by companies. Invariably out of date and massaged to fit the task at hand, what started as a partial look at a particular audience suddenly risks becoming wildly misleading ‘insights’ that could spell disaster for marketers that bet their careers on them. This sort of data manipulation should be avoided. Instead, marketers should make it their focus to turn over every rock in the search for owned data that is directly linked to customer buying decisions, for example, customer relationship (CRM) systems. These are more complete and up-to-date and will drive better business decisions.
Three myths about AI:
1. The robots are about to take over
While some might be concerned about robot dogs opening doors, (perhaps unnerving for some), we’re still some way off ceding power to our robot overlords. Guru Banavar, the head of the team at IBM responsible for creating Watson (the AI system that mastered Jeopardy) told Tech Republic that most people don't have a good understanding of what machine learning is. He said: "We can teach a computer to recognise a car, but we can't ask that same computer, 'How many wheels does that car have?' or, 'What kind of engine does it have?'. Can you ask anything else about what this car is made of or how it is made? None of those things is possible. Those are all far away."
The type of AI known as Artificial General Intelligence is still a long way off.
2. Machines will take all of our jobs
Not all, but they will do better at some than we ever can. Mundane, repetitive and dangerous jobs will be replaced. This is nothing new and has proven to be a good thing in the past when it comes to our general well-being and longevity, as has been the case as we’ve moved from the fields to the factories to the modern internet age.
The good news is that as some jobs are phased out, as repetitive and rote jobs are automated, new industries will emerge, which will create a host of new opportunities.
3. Machines will swallow all your personal data
It’s true that AI systems have the ability to collect, store and analyse vast amounts of consumer data – but so do non-AI systems. It’s important to note that the same privacy by design principles that should be applied to a traditional approach can and should be applied to AI-based systems (especially in light of the recent GDPR implementation in Europe and new regulations on the horizon in the US and around the world). Consumer data must be handled carefully and transparently. What’s more, companies need to ensure they’re gathering only the minimum amount of data they need to deliver the service consumers expect.
The good news for APAC marketers is that they are ahead of the global curve when it comes to adopting, understanding and embracing the benefits of AI, according to Adobe and Econsultancy’s Digital Intelligence Briefing: 2018 Digital Trends report. Those that are combining digital marketing skills with technology are nearly twice as likely to have surpassed their 2017 business goals by a significant margin (20% vs. 11%).
As data gets more abundant and processing get smarter, the innovation available to marketers will multiply. We’re already seeing it in the way in which consumers engage with brands today, but we can expect it to disrupt every consumer engagement, every company and every industry in the coming years. Armed with these myths and truths, you’ll be better equipped to spot the most trustworthy signals.
Konrad Feldman is founder and CEO of Quantcast.