Ahmed: Technology won’t eliminate the need for human scientists

Meet Adam. He’s a scientist who uses logic, formulates his own hypotheses, tests those… Meet Adam. He’s a scientist who uses logic, formulates his own hypotheses, tests those hypotheses and compiles results — all on his own. His hypotheses about the functions of certain yeast genes, for instance, derive from his own reading of the yeast genome.

Oh, and he’s a robot. Did I forget to mention that?

In a 2009 article in Science magazine, researchers presented their creation, Adam, as proof that science had become automated. The researchers gave this combined robot and computer the logical language necessary to reason on his own. They gave him all the known genetic information about yeast. They put yeast cells in his little freezer. Then, they informed him of 13 genes that scientists had not yet discovered a function for.

Of the 13 mysterious genes, Adam successfully determined the function of 12. The Science article details how the researchers used their own methodology to confirm Adam’s work.

What are the implications of this development? Will this technology deprive science of human agency? Is this yesteryear’s light microscope, or does a robot carrying out entire experiments mean a certain threshold has been breached, and that, at some point, scientists will become obsolete?

To gain better perspective on the issue, I spoke to Jeffrey Brodsky, Ph.D., a professor in Pitt’s Department of Biological Sciences. His laboratory uses yeast to study, among other things, the method by which the cell performs quality control on protein manufacturing.

The Pitt News: What do you make of this recent development? Does this mean science is becoming automated?

Brodsky: There is a generational trend. Each generation of scientists has looked at the previous and has been impressed by how hard it was and how easy it has become. For example, when I was a graduate student, it would take one month to sequence a genome. Now it never takes more than 24 hours. In the generation before mine, a graduate thesis could consist of purifying a restriction enzyme — an enzyme that cuts the DNA molecule at certain positions — and now, you simply order it from a company.

TPN: Does such rapid, automated technology take away from human agency in scientific research?

Brodsky: We are not letting go of that creative touch. Great moments in biology have come from people that have thought completely outside of the box. Sometimes, biology is illogical and so not suited for a computer’s reasoning. For example, a scientist, through a brilliant experiment, discovered jumping genes — genes that move around in a genome. No computer could have done this.

The computer uses mechanisms that are deductive and reductive. This results in the simplest conclusion — Occam’s Razor. The simplest conclusion is not always the case in biology. Biology doesn’t always reduce to the simplest explanation, nor does it always follow some sort of logic. The illogical, the surprises keep me getting up in the morning.

TPN: What do these sorts of developments mean for you?

Brodsky: It means people like me have to come up with hypotheses faster.

TPN: What do you foresee in the future?

Brodsky: I look forward to a day when I will become interested in a gene, and click a button to see if it appears in the genomes of 5,000 people with a disease, and in the genomes of 5,000 people without a disease. This can lead to personalized medicine. Also, such a database could allow people with hypotheses on the effects of a gene to easily test them.

Brodsky also noted the ability of humans to pick up on nuances. If a robot read a body of statistics and found a characteristic deviation from a mean, it would write it off as an error. A human might be intrigued if such a deviation appeared repeatedly. Perhaps it would be a clue to something overlooked.

Brodsky also mentioned that there are already computer programs meant to simulate doctors. Patients type in their symptoms and information, and the computer, using algorithms and a database of diseases and symptoms, churns out a diagnosis. It’s been demonstrated that such programs are much more accurate than real doctors’ diagnoses.

But again, there’s something missing: nuance. Doctors can usually tell when patients are lying. They can also tell if they’re hiding something — something typed text cannot convey. Sometimes a speech pattern or set of mannerisms implies a mental illness. Computers cannot yet pick these up.

I think we can answer some of my initial questions now. Certainly, such technology portends increased efficiency and accuracy in the future, but will computers replace scientists and doctors? It doesn’t seem likely. It’s human nature to be curious, see patterns and sometimes think illogically. It’s also in human nature to connect with other humans, help them through a diagnosis and understand them beyond a paragraph of symptoms.

Technology continues to advance science and medicine, but it will never replace humans in those disciplines — it’s just a generational tool for professionals to exploit.

Contact Abdul at aba24@pitt.edu