Natalia Lukaszewicz, Business Development Manager at CENSIS muses on how AI is becoming part of our daily lives
Since launching late last year, ChatGPT has already amassed over 100 million users from across the globe, making it one of the fastest-growing online applications in history. Most of us will have logged on and tested its capabilities in some form – largely out of curiosity – but, for many industries, the power of artificial intelligence (AI) could be transformational.
Ethical, economic and technology frameworks for AI – and the impact these could have on different end markets – were all discussed in depth at March’s Scottish AI Summit. The event took place in person and online, covering a range of topics from diversity and dialect of AI languages to workplace culture and the future of journalism. Sectors ranging from education to fintech are set to benefit from the new wave of automation, and sessions even looked at the power of AI in fighting cross-cutting themes like climate change.
Given ChatGPT’s current popularity, much of the conversation was about the opportunities presented by these kinds of large language models, however, speakers also recognised the potential limitations and risks of becoming overly reliant on these tools. After all, an AI system is only as good as the sum of its parts, and the information used as the starting point must be precise and unbiased.
The summit left participants with lots to think about and, for me, three big questions about AI and its role in work and life:
How to build responsible AI?
Toju Duke is a responsible AI programme manager at Google and in the keynote session raised interesting points about ethics and ensuring fair use of AI. Already, there have been instances where AI programmes have been built using biased information, seemingly causing more harm than good as a result of mistakes made.
How then do we ensure the reliability of data sources and the quality of the inputs into these programmes? Perhaps it is through a framework, like the one Toju suggested, where factors such as fairness, online safety, privacy and security are considered throughout the process. Equally important is the explainability of systems and being transparent with those who use or share data with them. Users need to know that they are interacting with an AI-based system, particularly if they are going to base decisions on that information.
Linked to this, we also heard about an interesting initiative between the Alan Turing Institute and Scottish AI Alliance, working with Members of the Children’s Parliament to understand their awareness and perceptions of AI. The programme covers both the technical and ethical sides, but on the whole the children had balanced views, being aware of AI’s limitations and risks, but positive about the potential benefits too. Again, this reiterated the importance of transparency and ensuring that the next generation understands their rights when it comes to AI.
Will AI become part of our day-to-day lives?
Although they are not always helpful, most people will have interacted with customer service chatbots before. But, there are plenty of other instances where AI programmes could be incredibly useful in day-to-day life.
One of the most interesting use cases of AI discussed as part of the two-day event, with a real-world benefit for many people, was in the healthcare sector. Precision medicine is a new approach to care which uses treatment and prevention tactics based on individual circumstances, rather than one-size-fits-all techniques. This relies on clinicians having access to vast amounts of data and information.
Using an AI tool to help with diagnostics and treatment recommendations could reduce the burden on clinical staff as well as providing valuable insights. Data accuracy and quality would be especially critical components of using AI in this setting.
In the coming months, CENSIS will be hosting a workshop with the University of Glasgow’s Living Laboratory for precision medicine, to explore similar opportunities where technology could unlock opportunities to develop healthcare services. Of course, there are useful applications of AI across many other industries, but this feels universal – at some point, every one of us will require medical assistance.
How will AI impact the future of IoT?
For IoT and sensing technology, AI is an incredibly valuable tool in terms of processing data. The more complex the system is, the more data generated in the system, which makes the scale of delivery immense. AI and ML (machine learning) enable automated actions based on the data output (a closed feedback loop). The system can be trained for self-intervention and autonomous self-organisation that does not require human intervention. Quickly developing AI and ML technologies push the boundaries and open even further the scope of application and functionalities, such as prognostics, condition monitoring.
Using AI could help to streamline systems involving thousands of sensors and data points – for example, in predictive maintenance or digital twins. Again, for it to work effectively, it needs to be precise – something which appears to be a common concern across each and every example.