Back when the buzzword of the day was Big Data, the following quote was all over conference presentations

Big data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it…” – Dan Ariely

Today, the same is true of Artificial Intelligence (AI).

AI will accelerate marketing and sales”, so said Dharmesh Shah of HubSpot at Inbound 2016. Zenith Media has predicted 10 AI trends for marketers for 2017.

What AI can do for you is being flaunted around quite openly, and everyone is telling you that you need AI now, or you are behind the curve.

As a marketer, when you evaluate how you bring in the benefits of AI to your company, do you chose to work with a partner organisation just because they say they have AI, or discount a company because they don’t have it?

Let’s peel back the layers and…

Understand how they define AI

At its simplest, AI is a computer driven decision based on a series of rules.  These rules can be:

  1. Defined by a human,
  2. Defined by the machine itself (ML or Machine Learning)
  3. Abstracted machine learning where it’s practically impossible to understand how the program makes its decisions (DL or Deep Learning).  

Quite often, these terms are used interchangeably.

It is important for marketers to understand that anyone who claims to have AI or is promising to give you benefits of AI, should be able to tell you how it works, how it was trained, how it was tested and how accurate it is.

If they can’t explain this to you clearly than you can’t be sure that you are going to get the correct results. Would you buy a car without knowing how many seats it had or how far it can go?

AI can give you powerful insights into your data but you need to understand what the tool is that you are using.

Is the AI ‘mutton dressed as lamb’?

Some people will try to package AI in different ways to sell to you. Some companies who say they have AI, actually don’t.

Some companies use a small amount of AI and then power them with human workers to get the final decisions or labels on the data. This is called machine-washing: pretending you have machine learning when it’s really hand-cranked behind the scenes.

If you’re sold something as AI when there is a human element then you will not only have a solution that is slower than one that is 100% AI, but you will also be at the mercy of human fallibilities: from local holidays and illnesses through to perception fatigue.

Anyone with AI will be able to give you a real time demonstration. Don’t get “washed”!

Understand their level of AI accuracy

Us humans are fallible. We get things wrong. After a day looking at images or numbers we can miss things, a momentary lapse of concentration and we click the wrong button. This is known as human level accuracy.  

AI is measured on specific tasks against how accurate an experienced person would be at the same task. A good AI needs to be faster and more accurate than a human equivalent.

This leads to two more terms that are often misused as ‘accuracy’; Precision and Recall.

Precision: the number of correct results from all the labelled results.  If you have a system that has 95% precision on matching your brand’s logo, then 5 out of every 100 labels will be wrong.

Recall: the number of correct results out of all possible correct results.  If you have a system that has 75% recall on matching your brand’s logo then the system will completely miss one out of every 4 logos.

Most AI systems are a balance between these two measures. Most big players don’t get 100% accuracy, so be wary of anyone who claims their AI is perfect.

As marketers you need to understand, which is more important to you?  

Without understanding this difference you could buy a system that claims to be 99% accurate when what is really meant is that it’s 99% precise but may have a very low recall rate.

Don’t let companies confuse you with the wrong terms.

If you’re interested in finding out how real AI can make a positive impact on your marketing ROI, get in touch with one of our experts today.

Get In Touch

Janet heads up the Artificial Intelligence division at StoryStream and has extensive experience in leading teams and building AI systems. With over 10 years experience at C-level, she has taken abstract concepts to innovative saleable products at multiple companies. A STEM polymath, Janet brings multidisciplinary ideas to the departments she leads, forming cutting edge research teams with an eye on delivery that is rarely seen outside of academia. Janet is a founder of the Tech Women London Meetup Group, Treasurer of the IEEE STEM strategy committee and regularly speaks and writes on various technical subjects.