I have always been interested in technology – I would drive my family crazy by taking things apart and putting them back together. I am fascinated by the promises of emerging technologies and get a real buzz out of stretching them to the limits of their capabilities, to test how robust they really are.

I began my journey in earnest in AI research in 2019. At this time, I was fairly confident that we would see the technology developed, tried, tested, and reaching the stage of acceptance and full implementation - as it has been in so many other avenues of our daily lives - before I completed my doctoral studies. As we all know, this has not come to fruition for many reasons. Some of these reasons are valid and enduring, and some require an adjustment of our human expectations and willingness to build a meaningful relationship with technology.

I think as imaging professionals, we appreciated there would be challenges and the road to successful AI implementation would never be smooth. We have learned from other technological advancements that impact human end users, there can be unpredictable and unforeseen complications, often resulting from the impact on human engagement. However, with modern forms of AI, the potholes in the road have been many: from AI applications with limited clinical use, perhaps before the clinician played an integral role in inception and development, to integration of these systems into practice where time requirements (more clicks!), trust issues and cognitive biases may also result in under-use of systems already in place.

Do we feel like we are in the car with children in the back seat shouting ‘Are we nearly there yet?’ Or is it more like some version of speed dating, where we are expected to build up a healthy relationship with a system (attractive as it may be) that we haven’t had the chance to get to know? There seems to be one infallible truth however, demonstrated not only in our professional lives, but in our day-to-day function – AI and advanced technologies are here, and they are here to stay. 

This is, in large part, driven by the promise of return on investment and user preference. 

The awareness of the public in the advances of technology is now being realised and we need to be ready to justify the inclusion, or indeed, exclusion, of technology in the patients’ pathway, something that some studies suggest we are not fully ready to provide. 

At the European Congress of Radiology (ECR), held in March in Vienna, Austria, I found myself, quite literally, lost in the AI expo. I spent 40 minutes trying to arrive at a scheduled meeting but, while my feet were aching, I was able to absorb some of the vastness of the field of AI in radiology as it currently stands. There were approximately 50 vendors, all with a slightly different perspective and offering, all trying to find their niche in the market. Unlike last year, many vendors were now ‘ready to go’ with the appropriate medical regulation approvals in place. However, acquiring the required approvals may be proving a challenge especially within the UK, where different approvals may be needed in the devolved nations of Northern Ireland and Scotland. 

At ECR there were different user interfaces and feedback formats demonstrated and tentative descriptions of how these would integrate into existing systems, however until the system is installed and used in the setting, this may be purely academic. Many vendors are fine-tuning the interfaces to suit the clinician or the clinical scenario to circumvent such issues. This is to be commended as we are discovering there is no ‘one size fits all’ to our relationship with AI. The implementation stage needs to be led by knowledgeable clinicians, who are aware of the needs of both sides of the human-AI relationship, in order to nurture and develop healthy engagement. 

With these many and varied options available, we need to be in a position to quickly engage with different forms of AI, but ultimately arrive at the solution which we want in order to build a long-lasting relationship with clinical AI.

In answer to the voice from the back seat – no, we are not there yet, but we must not passively watch the world go by. We know where we want to get to and we are going to choose the best route possible to get there, avoiding the hazards and potholes as best we can. 

Now is the time to get involved, to shape how the technology will be used. We need to build our relationship with the systems available to help us, but in a reciprocal way – there has to be give and take. We need to find out what we gain from the relationship, but we need to allow AI a seat at the table, on our terms. 

Read more