The Upcoming Unavoidable Challenge in Scientific Research
Written on
In the realm of art, Leonardo Da Vinci is often celebrated as a master painter and a pivotal figure of the Renaissance. However, he was also an architect, engineer, inventor, and, most significantly, a scientist. Born in 1452, Da Vinci explored the world systematically, designing steam-powered cannons and flying machines, experimenting with early optics like the camera obscura, and producing intricate anatomical illustrations through dissections. His grasp of contemporary knowledge was profound, often surpassing that of his peers by a century in fields such as friction physics. This extensive knowledge enabled him to integrate his discoveries into groundbreaking inventions, like the mechanical knight.
To achieve a similar level of understanding today, one would need to be akin to a robot—or more accurately, an advanced AI that we have yet to develop. The sheer volume of scientific literature published provides insight into this complexity.
2.5 Million
In 2015, the scientific community published 2.5 million articles, averaging about 6,850 daily, or roughly five every minute. Each article typically took around 30 minutes to read, meaning that by the time one finishes reading a single paper, 141 new ones have already been released. The increase in published articles is estimated at 4% annually, with some sources suggesting even higher rates. This translates to an additional 100,000 papers each year, or about 273 more per day. This figure exists atop the already staggering 2.5 million publications.
Some critics argue that many articles are published for trivial reasons, leading to wasted time, while many others are not published, depriving scientists of crucial information. Currently, there are approximately 7.8 million researchers globally. With 2.5 million annual publications, this means that, on average, each researcher publishes just one paper every three years. Even if each focused solely on one topic throughout their 40-year careers, they would collectively produce around 195,000 papers annually—one every three minutes. Thus, it's clear that no individual can hope to read everything being published.
Why Does This Matter?
You may wonder why this is significant. We have specialists who keep abreast of their fields, right? Why does it matter that no one can stay updated on everything?
Even highly specialized topics can generate an unmanageable number of papers. For instance, in 2020 alone, there were 87,000 papers on the Coronavirus—one every six minutes. The crux of the issue is that all fields are interconnected. For example, if an entomologist studies a butterfly capable of navigating thousands of kilometers, they face challenges in determining how it does so. While the butterfly might use its eyes to gauge sunlight, this raises questions about its internal sense of time and navigation methods. Despite lacking magnetic materials, the butterfly's compass relies on light-sensitive quantum effects.
A researcher focused solely on entomology might struggle to understand these complex interactions without knowledge of quantum chemistry. If they have access to existing literature, they might find connections rather than starting from scratch, but someone had to establish this relationship first. Not all disciplines have such obvious links, as demonstrated by a friend researching neuron signals with an algorithm designed for RADAR technology. If he hadn't been open to interdisciplinary connections, his research would not have succeeded.
Discoveries often arise from intuitive leaps that recognize unexpected links between seemingly unrelated concepts. Many scientists engage in multiple fields, as a study of Italian professors revealed that only one among them focused exclusively on a single subject. The rest published across various areas, often only loosely related.
Yet, with the overwhelming influx of information, keeping up with developments across all fields is nearly impossible, leading to frustration for those eager to comprehend the universe.
We Lack Awareness of Our Knowledge Limits
This frustration is felt by both scientists and the general public. Numerous discussions online indicate that we are nearing the limits of our understanding due to the sheer volume of information available. The total number of unique scientific papers remains uncertain, with estimates around 50 million, but the true figure is elusive—a concerning sign.
One requirement for doctoral theses is that they must contribute original knowledge. Yet, how can one meet this standard when it’s challenging to ascertain how many papers exist in their field? Accessing relevant literature can be daunting, especially with paywalls or lost publications. Researchers might inadvertently replicate studies due to lack of access to prior work, or they might not discover simultaneous research occurring elsewhere. This lack of oversight hinders scientific progress, as many potential discoveries remain undiscovered due to scattered information.
Google as a Solution?
Search engines like Google Scholar offer some relief by indexing publicly available papers and helping researchers find relevant studies. However, these tools do not address issues like inaccessible articles, unpublished negative results, or duplicate research. Moreover, indexing does not equate to a comprehensive understanding of a subject. Without prior knowledge, researchers cannot make intuitive leaps that lead to new insights.
Currently, we lack the AI capable of solving this challenge. Existing AI methodologies, whether deep learning or symbolic programming, are not equipped for the intuitive leaps needed to connect disparate pieces of knowledge. While AI can assist by modeling known phenomena, it cannot identify or connect the myriad of papers and insights that exist. Future AI might bridge this gap, but until then, the sheer volume of data remains an obstacle, and machines cannot provide answers to questions we aren’t even aware we need to ask.
