top of page

MBSE powered by Artificial Intelligence will become the Systems Engineering of the future

There’s a revolution coming.

You may not know it, but you’re certainly seeing it.

Artificial intelligence is here. The most notable instance is Chat GPT3. A language model built using large datasets and reinforcement learning.

Model Based Systems Engineering (MBSE) is also here. Not yet widely adopted, but most companies are striving to build a ‘digital twin’ of the system before even thinking about starting to build anything.

We’re living in a time of huge leaps of computing capability, experiencing a drive to build ‘artificial intelligence’ which automates mundane tasking and an ethos of building a full representation of a system in a digital model.

The fire is primed for a revolution in Systems Engineering. All the key fuel is in place, we just need a spark.

The spark which forges MBSE powered by Artificial Intelligence.

Here's how it's going to happen….

First, it’s going to begin with the automation of the mundane tasks when building a digital model.

The linking of requirements, the dragging of blocks onto diagrams. The creation of sequence diagrams that have been created a million times over.

Once sufficiently capable at such tasks it’s going to evolve into a recommender system.

It means that a neural network model, trained on vast quantities of previous project datasets will ‘recommend’ a proposed system for you to use to fulfil your requirements.

Now, it's worth stating that we are a million miles away from this.


Three main things;

  1. Datasets

  2. The need

  3. Modelling capability

Simply put, we don’t have the data set to do it successfully. We may have some data, but not enough to build entire systems. The building of the dataset is a mammoth undertaking. Likely to take a team years to complete.

Furthermore, we don't have a need for it. As ‘cool’ as it may sound, it's a lot of investment for something that is ‘nice to have’.

Finally, we’re not there yet in terms of modelling capability.

Chat GPT3 cost an estimated 12 million dollars to train. Furthermore, it costs about $100,000 a day to run.

It’s super expensive, who in the Systems Engineering landscape is going to spend the guts of 20 million to get a systems recommender off the ground.

In a more optimistic outlook, there are examples of ‘code recommenders’ which is the start of this modelling revolution.

However, they are pretty ‘cookie cutter’ examples when you talk to any engineer worth their salt about them.

We would need a vast enhancement of computing power to reduce the cost of undertaking to make it even remotely considerable as an investment of time and resources.

All of the above will happen. Probably not in the next 5 years, but likely in the next 10-15.

To leave you on a cliffhanger…

How much of the ‘boring’ parts of your job could be replaced by a model?

Picture the following tasks just to start!

  1. Reviewing Technical Standards

  2. Writing requirements

  3. Proposing changes to requirements

  4. Formulating architectures to fulfil requirements

  5. Designing components configuration that have already been designed a billion times over

  6. Analysing performance of your designed system

  7. Verification testing!

  8. Safety case builds

  9. Operator manuals

The list is ENDLESS!

We get excited even thinking about it.

No need to stress about your job security.

You won’t be replaced by such a job.

Actually, the model would augment your work by ‘recommending’ routes to make your life easier and fulfil your mission.

Then you’d transition into reviewing the outputs of the model and verifying the suggested results.

A modelling revolution is coming! Just not yet…

Recent Posts

See All


bottom of page