Call us now:
Thanks for posting Dan Lohrmann. Cyber-enabled economic warfare is not new, Not Petya was an example where cyber got out of control and effected economic stability and we are seeing more of it. But to answer the question about AI and cyber-economic warfare, you have to understand the main types of AI, that include Machine Learning (ML) and language based models. When one goes back 25 years developing ML systems for the automation of the Investment Casting process for Turbine Blade manufacturing. We worked on two simple yet complex capabilities that Machine learning AI needed to function. The first was to be able to correlate the relationship between Cause-Event and Impact and the second was to manage variation. With the aim of building ML models that would repeatably correlate to the real world.
Machine Learning and Language based models require vast amounts of data. They rely on being able to correlate data inputs and outputs through either probability based models, AKA the reasonable probability of getting a repeatable output response, when input responses change through understanding causation. Or being informed of the relationship between what goes into a model and what comes out based upon what they know and have been told (language models).
Building ML models for investment casting back in the 1990s, the models needed to understand the correlation between complex manufacturing variables (inputs), processes functions and outputs (blade geometry). Models that would change as soon as inputs changed through variation. Variation that took place both intentionally (we changed things) and through natural variation (sh1t happens). At the time we found that it was not possible to repeatably correlate inputs, process with outputs due to complexity and variation.
How does this correlate to cyber? Firstly we do not know the relationships between Cause, Event and Impact (causation); Second the process to manage cyber security is very complex for example we do not understand how changes in cyber variable effect each other and outputs, and thirdly there’s so much variation in inputs, the process and outputs it makes correlation difficult.
However, that does not discount AI, as it can be applied to discrete cyber processes, demonstrated in the article as its applied to the specific process of a cyber attack, where causation is understood and can be modelled. The challenge comes when it is applied to defensive cyber that is more complex.
Will AI handle cyber-economic warfare, not yet.
Ted Dziekanowski
#cybersecurity #cyberriskmanagement #cyberattack #ai
View Original Post
https://www.linkedin.com/feed/update/urn%3Ali%3Ashare%3A7080981591560212480