By Marjolijn Bijlefeld, with an assist from AI
In our hyper-connected world, we can do a Google search on almost anything. The challenge is phrasing the question so that the answers direct me to the most appropriate rabbit hole.
We’ve all been reading about the potential upsides of artificial intelligence (AI) in optometry. These range from improved outcomes through faster and better analysis of data, increased efficiency and improved decision-making, applications in remote monitoring and earlier prediction of patterns to detect eye disease.
So how is that different from what optometrists are already getting through the data in normative databases? These normative databases have been created for so many conditions, from myopia to age-related macular degeneration and glaucoma progression and more. I asked AI to help me out.
TWO LESSONS
Be brief. I asked for a 500-word answer to describe the differences between normative databases and AI algorithms. Big mistake. AI gets repetitive, like a kid who only read the first chapter of a book and needs to turn in a two-page report. I kept half of the AI answer below. Next time, I’ll ask for a shorter response, and I can follow up.
Be specific, but know that AI might not have an answer for you today. When I asked a follow-up question related specifically to normative databases and AI algorithms in eye care/optometry, my AI gremlin apparently went on a coffee break. When i tried some different phrasing, AI threw the word “optometry” into the same basic response it had given me earlier.
THE HUMAN-EDITED AI ANSWER
One of the key advantages of AI over normative databases is its ability to adapt to new data. While normative databases are based on fixed data sets, AI algorithms are designed to learn from new data as it becomes available. This means that AI can adapt to new situations and make predictions based on the most up-to-date information available.
PROS
AI can identify complex patterns and relationships in data. Normative databases are typically limited to basic statistical analysis. AI algorithms can identify correlations and relationships that might not be immediately apparent to a human analyst. For example, an AI system analyzing data on patient outcomes might be able to identify a correlation between a particular treatment and a lower rate of hospital readmissions, even if that correlation is not immediately obvious from the raw data.
LIMITATIONS
One of the key challenges with AI is ensuring that the algorithms are accurate and reliable. The algorithms are only as good as the data they are trained on. If the data is biased or incomplete, then the AI algorithm may make inaccurate predictions.
Also, AI lacks transparency. It can be difficult to understand how the complex algorithm arrived at a particular prediction.
NOT TOO BAD?
What I wanted to know was how AI algorithms and normative databases differed. The AI response gave me a fairly succinct response that summarized key points. When I did a Google search on the same question, Google invited me to review 920,000 results. Who’s got time for that?
How are you using AI in your practice or research or teaching? Email me.
Read opinions from other readers here.
Featured photo credit: Getty Images, Moor Studio
