Israel is the leader in forward technology adoption and capabilities of AI is not overlooked. Deploying AI across all branches of the military positioned Israel away from Human Intelligence and towards an automated data analysis model. The results are in. AI looks to have failed yet once again.
In 2022 Israel declared a new military strategy based on Artificial Intelligence, https://www.c4isrnet.com/artificial-intelligence/2022/02/11/israel-unveils-artificial-intelligence-strategy-for-armed-forces/ . Although the exact details of where AI solutions are deployed are unknown a fair guess would be video surveillance and human detections sensors. Who doesn’t have a web cam at home that detects people? AI is known to be deployed with law enforcement and such solutions generally connect people, places and things so we can probably assume similar solutions are deployed with the Israeli military, https://www.ibm.com/docs/en/announcements/archive/ENUSA16-0081 . In summary most of Israeli’s threat analysis and awareness may be driven by AI. Humans may be collecting the information but not analyzing it.
On October 7, 2023 a group of opposing forces attacked Israel with seemingly little resistance. The attackers, rode mopeds across the border, flew in with para gliders and often just walked in without sophisticated electronics measures. Headlines read about failed intel, lack of preparation but no mention was given to Artificial Intelligence. From the various online reports, the opposing force looks to have penetrated all the areas that would have been monitored by AI solutions. So, was what happened another AI failure?
Let’s look at known AI failures. Zillow bet a large amount of money on its AI solution that nearly destroyed the company, https://towardsdatascience.com/zillow-collapse-proves-ai-is-not-perfect-let-me-explain-3bdeb03b7713. Tesla uses AI to drive its auto pilot and still continues to make the news in a negative light, https://www.caranddriver.com/news/a44185487/report-tesla-autopilot-crashes-since-2019/. Examples are abundant and follow a similar pattern. AI is deployed in critical processes and fails to deliver.
Can a poorly equipped opposing force defeated AI? Residence have defeated auto pilot by placing cones on the hood of self-driving cars, https://www.dailymail.co.uk/news/article-12277917/Activists-putting-cones-hoods-self-driving-cars-stop-San-Francisco.html. Facial recognition AI solutions were consistently defeated by people wear face masks, https://cyber-reports.com/2022/07/12/researchers-defeat-facial-recognition-systems-with-universal-face-mask/ . Clearly, poorly equipped opponents with ingenuity can defeat AI.
What happened on Oct 7 shocked the world. A country known to be a champion of AI was betrayed by it. AI failed to protect Israel. The question now is how does a country un-AI itself and go back to the processes AI was intended to take over?
Our thoughts are that going forward organizations will need to develop AI fail-over systems to mitigate risks which have already been architected into the organization.