A major stumbling block in our understanding of age-related disease, such as Alzheimer’s, is a propensity to focus on large numbers of genes, proteins, etc., without asking what lies “upstream” that results in the associations between such genes (etc.) and the disease. While some would tout the advantages of using Artificial Intelligence to attack the […]
The End Hangs on the Beginning
A major stumbling block in our understanding of age-related disease, such as Alzheimer’s, is a propensity to focus on large numbers of genes, proteins, etc., without asking what lies “upstream” that results in the associations between such genes (etc.) and the disease. While some would tout the advantages of using Artificial Intelligence to attack the problem, the problem with AI is that (like most scientists) it focuses on finding solutions only once the problem has been defined ahead of time. If, for example, we define Alzheimer’s as a genetic disease, then we will find genes, but will never reassess our unexamined assumption that AD is genetic. If we assume that it’s genetic, then we only look at genes. If we assume it’s due to proteins, then we only look at proteins. If we assume that it’s environmental, then we only look at the environment. Data analysis and AI, no matter how powerful, is limited by our assumptions. We tend to use large data analysis (and AI) in the same mode: without ever realizing we have narrowed our search, we assume that a disease is genetic and then accumulate and analyze huge amounts of data on gene associations. While AI can do this more efficiently than human scientists, the answers will always remain futile if we have the wrong question. Once we make assumptions as to the cause, we only look where our assumptions direct us. If we look in the wrong place, then money and effort won’t correct our unexamined assumptions and certainly won’t result in cures.
It’s like asking “which demons caused plague in Europe in the middle ages”? If we assume that the plague was caused by demons, then we will never (no matter how hard-working the researcher, how large the data set we crunch, or how powerful the AI we use) discover that the plague was caused by a bacteria (Yersinia pestis). If you look for demons, you don’t find bacteria. If you look for genes, you don’t find senescent changes in gene expression. The ability to find answers is not merely limited by how hard we work or how large the data sample, but severely and unavoidably limited by how we phrase our questions. We will never get anywhere if we start off in the wrong direction.
To quote the Latin phrase, “Finis origine pendet“. The end hangs on the beginning.
This is wisdom not only for scientific research, but for life in general!