Last month, I had the pleasure of attending the ‘AI for Critical Systems’ conference (CWTEC) 2024, hosted by Cambridge University Computer Laboratory. With a record-breaking 160 attendees, it was clear that AI is becoming a big deal in critical systems like telecommunications, cybersecurity, healthcare, and banking.
Key Takeaways:
The day kicked off with an engaging introduction to AI by CEO Michaela Eschbach, who brilliantly covered AI fundamentals and recent breakthroughs, setting the stage for understanding AI’s role in critical infrastructure. Guest speakers delved into how AI can enhance diagnostics, resilience, security, and efficiency across critical industries. It was revealed that AI is predominantly being utilised for optimisation, troubleshooting and anomaly detection purposes.
Professor Amanda Prorok from the University of Cambridge delivered a standout keynote on deploying AI in multi-robot systems. Her studies showed how machine learning enables robots to exhibit coordinated and cooperative behaviours, and in one example, can enable formation control of robots. I couldn’t help but wonder if these robots could teach the recent England Euro’s football team a thing or two about working cohesively!
Telecommunications and Transport: Chris Murphy explained how AI is aiding in the creation and maintenance of resilient networks, by using digital twins. These virtual models of a network help service providers manage network behaviours and disruptions more efficiently, making real networks more resilient.
Cybersecurity and Bioinformatics: AI can help detect potential issues and fill in missing data points and also build on existing data science approaches in bioinformatics. The availability of large datasets is crucial for these advancements, though heuristics and human oversight remains essential for robustness.
Challenges and Concerns
But let’s not get too carried away—AI isn’t all sunshine and rainbows. There are some significant challenges:
- Transparency and Security: Black-box models can produce hallucinations and are vulnerable to malicious actors. Not ideal for critical systems!
- Skill Fade: Automation risks eroding human skills, making it tough for people to step in quickly if something goes wrong.
- Privacy: Using personally identifiable information and biometrics, such as facial recognition, requires significant caution.
- Energy Demand: The increasing energy consumption of AI poses climate risks, potentially necessitating dedicated nuclear plants for data centres.
Future Trends
Despite the challenges, there’s a lot to look forward to:
- Advanced Architectures: Enhancements like neuro-symbolic AI and improvements in transformer architecture are boosting AI’s capabilities.
- Innovative Approaches: Generative Flow Networks (GFlowNets), proposed by Yoshua Bengio, and neuromorphic computing, exemplified by The Thousand Brains Project, offer promising directions for the future.
Why It Matters
As members of Cambridge Wireless, attending CWTEC 2024 was brilliant. I gained crucial insights, explored disruptive technologies, and enjoyed excellent networking opportunities. These events are essential for staying at the cutting edge of AI advancements and understanding their practical applications in critical systems.
Overall, the conference was a valuable forum for sharing ideas, exploring disruptive technologies and forging relationships – bringing together bright minds to discuss the potential and practicalities of implementing AI solutions that enhance the security, efficiency, and performance of critical infrastructure. While AI holds significant promise for many industries, it is clear that there is still a long way to go to fully realise its potential.
