Every media outlet in the land loves a good AI story. There are loads and loads of AI myths floating around. Ranging from ‘The robots are coming for our jobs!’ to ‘The robots will usher in a Utopian society!’ and everything in between.
The truth, as always, is not quite as drastic.
As AI is becoming increasingly prevalent within marketing, let’s take a look at some of the most common AI myths out there, and lay some misconceptions to rest:
Myth 1: Feeding data to a machine will make it smarter
The number-crunching aspect of digital tech is very appealing to many marketers. Some of the most monotonous and repetitive work in marketing involves data.
Machines are capable of gathering, processing, analysing and utilising data constantly. They’re incredibly efficient, and they never get bored.
However, there are machines and machines.
Some machines can (sort of) ‘learn’ from data and grow ‘smarter’. These are Artificially Intelligent machines, capable of spotting data patterns which you can use to predict things like customer behavior.
But not all machines are AI.
Feeding data to an automation isn’t going to have any effect (beyond possibly slowing it down). And, even those machines which are AIs don’t exactly become ‘smarter’ in the traditional sense the more data you feed them.
Explaining this leads us neatly into Myth 2…
Myth 2: AI and Automation are the same thing
AI, Machine Learning, and Automation are terms which get jumbled about a lot. Let’s break it down:
- AI and Machine Learning are closely linked. An AI is a program, and Machine Learning is a tool that program uses to improve its functioning through experience.
Machine Learning is what helps the AI at Amazon to predict the kinds of products you might like based on your past purchases.
AI is more of an umbrella term, covering everything from that Amazon bot to neural nets designed to mimic human cognition being tested in futuristic labs.
- Automation is a process whereby a machine or program will do things automatically. It can be informed by and integrated with AI, but an automation is not inherently AI.
Take a list-cleaning automation, for example.
You can program it to remove inactive or fake entries, and it will do that endlessly, forever. But it won’t learn from what it’s doing.
A list-cleaning AI, on the other hand, is capable of ‘spotting’ patterns in inactive user data, and offering insights about them.
Myth 3: AI will make me redundant!
AI will not make you redundant.
What AI will do is take away the less fulfilling parts of your job, and make you tons better at the more fulfilling parts.
Far from booting marketers out onto the street, AI to date has actually created lots of new roles. From data scientists to Chief Information Officers, AI has opened up a world of employment possibilities.
Essentially, AI gives human workers a leg-up. By taking on the time-consuming and mundane tasks of data gathering, entry, analysis etc, it boosts humans up to the next rung – enabling them to devote their time to work more suited to our unique human skills. Creative content work, for example, or relationship-building, or innovative campaign development.
This myth is so inaccurate that the World Economic Forum predicts that, by 2022, AI will actually have generated almost 60 million more jobs than it replaces.
So, you can relax. The robots aren’t coming for your job.
Myth 4: AI can work on its own
‘Clever’ as they are, AI programs don’t have autonomy or initiative. They will do what you tell them, and do it extremely well, but they won’t do anything you don’t explicitly program in.
Let’s say, for example, that you want to write some great email subject lines. Your AI is theoretically capable of running the numbers and coming up with something that’s likely to dazzle your customers the second your email hits the inbox. But it’s not smart enough to do that all by itself – no matter how much subject line data you feed it.
It’s up to you to guide it. You need to tell it things like the audience you’re going for, the effect you want, the keywords which need including, and so on.
AI is smart, and feeding it data will technically make it ‘smarter’. But only in the very specific, very prescribed ways that you tell it to.
If you program an AI to write subject lines, it’s not going to one day have devoured so many subject lines that it’ll quit your team and embark on a career as a novelist. That’s not how it works.
Myth 5: AI and Machine Learning always give accurate results
When they’re programmed right, AIs can give very impressive results.
But they’re far from infallible.
Any AI is only as good as its programming. Program it poorly, and it will give poor results.
It’s also worth remembering that – while they can be surprisingly insightful – even the best AI results lack the human nuance that makes them actionable.
An AI is a tool. It’s not an oracle. It’s a great tool, for sure. But it’s useless without a human to use it.