Accessability Links

Good AI Is like building a fire

Monday 10th February 2020

In this guest post we are joined by Chris Coates. Chris is talented DevOps professional who has over 10 years experience in IT.

He is not only very knowledgeable and experienced with using the latest in DevOps and AI technologies but is a very well decorated Rugby League coach, representing at international levels.

Here he talks about Artificial Intelligence and the tools and processes needed to get it right. Check it out below...

As someone who works with AI, I regularly hear that:

"Artificial Intelligence is the future"


"AI can do anything"

But often customer(s), sellers, and even engineers struggle to really get a good project off the ground.

So i'd like to suggest an analogy that is (sometimes overly) simplified, in order to help.

In my opinion, Artificial Intelligence is like building a fire. Yes, I understand this seems a little odd, and in some cases is oversimplification but let's go have a quick look at the "Fire Triangle" to understand where i'm coming from:

The triangle of Fire (lifted from Wikipedia)

Now - as you can see, to make a fire it generally needs 3 items - Fuel, Heat, and Oxygen.

The removal of any one of these things means the fire goes out, or cannot start.

Let's amend this a little bit, in order to understand a good AI project:

The AI "fire" triangle
So. Let's discuss this a bit:


The data is the fuel for the project. If you don't have enough of it? Your project won't last long.

No, i'm not saying burn the data!!!!

If you have the wrong fuel? Then the project won't work very well (just like a fire if you attempted to burn a stone - that won't work).

If you have poor quality data? Then your project will work, but poorly.

Take for example if you have some source data that needs cleansing because there are lots of erroneous results. This is often the equivalent of having "wet wood" in that it will work, but not nearly as well as if you had dry wood. The process of cleaning data is the same as drying out wood in that respect.

However sometimes no matter how much you dry it, it may just be the wrong fuel and that is something to be very aware of.

Get the right fuel for the fire? Get the right data for the project? And it'll work well.


Proper tooling is always important. The right tool for the job, just as in using the right algorithms and training model is critically important for an accurate AI project.

Without these tools? Just as a fire, without oxygen it won't work.

The wrong tool? Is like using standard open-air to burn a fire - but a good tool is like applying pure oxygen directly to the fire.

Make sure you're using the right tools for the job. If you're unsure? Engage the right people (shameless plug - I work for OCF Plc - come talk to me!) to help you decide.


Now, the correct tools and some good data? Is nothing without a spark to begin the project with!

Just as a fire needs that heat source to start the fire? A good AI project needs an idea or a problem in order to solve.

Lots of people expect AI to just be able to "solve it" when they don't necessarily know what they want to solve.

You might find that you have lots of data, but the problem you want to solve? Doesn't have the data that you need collecting.

Putting it all together

The process of putting all this together can sometimes just be a case of asking those 3 questions:

"What problems do you have in your day-to-day?"

"What data do you have that supports that you're monitoring those problems?"

"What tool is best to help solve this problem?"

You might find that during the process of putting this all together, you find that in fact you don't actually collect the data you need to operate on... Or (and this is common) that you aren't aware that you have the data. Data is everywhere! It's whether you collect it or have the tooling/storage to collect it.

A quick (and dirty) example
Take for example a guy in a bar says "I design cupboards every day - but some of them have quality issues when they're manufactured. I wish I could solve that" - Well technically, you probably could.

The first question would be "What do you do to track the faults?" "Do you track where the faults occur?" "Do you track it using something that can store the results in a searchable manner?" (for instance, video or pictures) "Does your data that you track have enough quality to operate on?" (i.e. high-definition imagery).

If you do? Then you already have your problem sorted. The problem is that the designs need to be amended or materials need changing, but the designer is unsure where. You have the data, because you can work out where the faults occur in manufacture using imagery... And you have a subject-matter expert (the designer and quality engineer) who can explain what the fault looks like.

Apply a good tool that can use this data, classify images and areas of images that identify damage and what it looks like?

And you likely have something that can absolutely run analytics on where the faults occur, and how the faults occur with a high confidence rate.

Then the person in question can leverage this information to change the design or the materials in those areas of common damage, and prove via the process of A/B tests using the same model whether that different design is better, or worse based on yield... Or it may be a trigger to go look at the machining that builds those parts and you could find a fault.

Problem solved!

Check out more of Chris' blogs by visiting his linkedin page here.
Add new comment
Back to Top
f24("cookieAnonymous", true);