The Medical Futurist – Read More
In the current age of artificial intelligence (AI), physicians are able to rely on the technology for a myriad of purposes. From triaging patients to assistance with radiological scans, they can turn to AI tools to assist them in their healthcare-related tasks.
However, by offloading cognitive tasks to automated systems, a new concern has arisen: are doctors losing their clinical skills due to AI? In this article, we reflect on the cognitive offloading phenomenon, its impact on healthcare practice and how clinicians can engage with AI tools without compromising their skills.
What is cognitive offloading?
In general terms, cognitive offloading involves the use of external tools to minimise the mental demands of conducting a task. A classic example is that of using calculators to carry out arithmetic tasks. By using this external tool, we are outsourcing the mental or cognitive need for our brains to do so by itself.
This process can be applied to a number of tools and, in particular, AI. With this technology, we can offload tasks such as fixing software code to writing whole books. Offloading cognitive tasks to AI is an attractive prospect as it offers a low-effort avenue to carry out complex tasks.
As with any tool, such task delegation comes with its own benefits and risks. Going back to the calculator analogy, using that tool can help make repetitive calculations easy while a student can rely on more abstract derivations. However, over-reliance on the calculator can also reduce the student’s ability to carry out arithmetic tasks for everyday purposes.
With cognitive offloading to AI in the healthcare space, the stakes are heightened. The technology is still relatively new, and we need to think about what happens long term if AI takes thinking out of physicians’ jobs for months and years.
Cognitive offloading to AI in healthcare: benefits and risks
Let’s first contemplate the benefits of physicians employing AI for cognitive offloading. In the clinical realm, retrieving and updating medical insights has traditionally been slow and fragmented. Physicians have to navigate through outdated platforms that take away their precious time. AI-driven systems such as information retrieval tools and automated scribes can handle cognitive tasks to prevent overloading physicians.
Indeed, some time-consuming tasks such as typing patient encounters in electronic systems can lead to cognitive overload for physicians. In turn, these contribute to physician burnout while making care delivery inefficient. AI systems can positively support physicians in routine tasks and improve the care experience.
However, there is a delicate balance that needs to be struck with cognitive offloading to AI tools. Becoming too dependent on these can make physicians reduce their autonomy in decision-making and minimise introspection, skills which are key to clinicians.
Studies indicate that doctors who rely heavily on AI output might have their clinical performance negatively impacted. One illustrative example found that endoscopists who regularly use AI assistance experience a notable decline in detecting cancerous lesions (from 29% to 22%) during subsequent non-AI procedures.
This raises the spectre of “cognitive surrender”, the more ominous-sounding consequence of cognitive offloading. As the term implies, this involves fully offloading cognitive tasks to AI, or surrendering all thinking to the tool. In turn, the user can falsely attribute the output of these tasks to their own judgement, when, in fact, it is the AI’s.
It’s a real risk that physicians can succumb to. In particular, for medical students and trainees, these concerns are more relevant. If the next generation of physicians over-rely on AI assistance without critically engaging with their work, they will fail to develop the independent problem-solving and diagnostic reasoning skills required for practising.
Can the risks of AI cognitive offloading be mitigated?
As dire as the consequences of AI cognitive offloading might sound, the practice does not mean a less skilled future clinical workforce. Mitigating the risks also does not rely on a wholesale ban of AI in healthcare. Instead, a more nuanced and critical approach is required.
If a cognitive task can be fully offloaded to AI – such as outright determining a diagnosis – its usage is likely inappropriate, especially for medical students and trainees. But if the tool can support learners in better engaging with the cognitive task at hand and take over more trivial tasks, then AI can be an adequate complement.
For example, when analysing a radiological scan, an automated assistance tool could indicate the region of a suspicious region and engage the physician in determining the reason for this indication. The latter could even challenge the AI’s thinking, as these tools are not impervious to mistakes.
As such, AI cognitive offloading would ideally require thoughtful design of the tool, robust policy safeguards, and careful integration. This is no simple task and will require concerted efforts to implement.
In the meantime, individual physicians can take responsibility to ensure that AI empowers their cognitive tasks rather than erodes their cognitive autonomy by critically engaging with the technology.
Written by Dr. Bertalan Meskó & Dr. Pranavsingh Dhunnoo
The post Are Physicians Losing Skills Due To AI? What Is Cognitive Offloading? appeared first on The Medical Futurist.
