AI and its Effects on Scientific Data

    Research and scientific data can be extremely valuable and sensitive to outside influences. Even a simple bias can wildly skew data in the incorrect direction, leading to false conclusions and untrustworthy results. AI has dozens of uses in scientific fields, though those uses shift based on the field and scientist using the machine. In the subdisciplines of environmental sciences, you are more commonly going to find AI used for tasks such as “air and water quality monitoring, climate change modeling, biodiversity assessment, and disaster management” (Alotaibi, Nassif 2024). On the other hand, in fields that relate to human health and well-being, you would be more likely to find AI completing tasks related to organic chemistry, catalysis research, and protein modeling. AI has wide and varied uses in the plethora of scientific fields and disciplines that simply have not been discovered or even thought about before. The existence of AI should be to make the lives of humans easier and happier, and scientific and medical discoveries are a large part of that. Learning more about the world and universe around us helps us protect ourselves and treat our planet in the best ways possible, ensuring that we will continue to thrive for generations to come. Knowledge of the human body, its systems, and their interactions will aid not just specific people, but humanity at large. Being able to cure previously terminal and fast-acting diseases and health conditions will improve the lives of millions of humans around the world. Research is already underway on this subject, and conditions that have so far had no cure or way to mitigate the damage done are having cures found and discovered every day. Additionally, “using AI in the preclinical stage of drug development could bring a time and cost saving of twenty-five to fifty percent” (The Economist 2024). However, AI may be used for negative purposes, and there are sound and valid security and ethical concerns related to its usage in scientific data, but we should not let those outweigh the good it can do for all of us.

Technology Overview:

    AI has incredible features and potential to be a force of truly good change on Earth. It has many kinks and issues that need to be solved, and fast, but that does not mean it is wholly evil like the movies may tell us. AI is an extremely versatile tool that has dozens upon dozens, if not even upwards of millions of applications and uses in the modern world. However, a major use of AI has been in the fields of scientific data. Fields such as environmental monitoring use AI and other machine learning programs and models to aid them. AI is used by these scientists in numerous ways, “Key areas of research include air and water quality monitoring, climate change modeling, biodiversity assessment, and disaster management” (Alotaibi & Nassif 2024). AI can analyze and interpret data at speeds several orders of magnitude faster than the human brain is able to. Making AI an amazing tool to help sift through and categorize data, as well as apply it to other projects or works such as the monitoring, modeling, and assessment previously mentioned. However, AI is not only useful for environmental scientists. It has also found a home within the fields of medical sciences. “These algorithms are being used to develop new drugs for diseases that are, right now, incurable” (The Economist 2024). AI is able to formulate and model proteins and molecules based on their components and simulate what they are able to do as their functions. Which allows scientists and medical professionals to experiment with and create new proteins and medical treatments much faster, and all without having to subject a living animal to the preliminary stages of development. Unfortunately, though, AI is not perfect, in fact it is far from it. AI is more than capable of providing false responses and untrue claims, these are called AI hallucinations. This is why it is always important to fact check and proofread what an AI program outputs for you, to ensure it is accurate and logically sound. Though, there are even more subjects that need consideration when it comes to AI. There are countless concerns that currently exist when it comes to AI. Many that have yet to be addressed in the slightest. But they must be if AI truly will have a cemented place in our society, at least, one that truly brings almost entirely positive change.

Legal, Ethical, and Policy Considerations:

    AI can be incredibly useful for a multitude of reasons. On the other hand, it is also somewhat unethical depending on the lens used to view it. Some would consider data that affects their health, or their personal research being put almost solely in the metaphorical hands of a machine to be dehumanizing or degrading, making them uncomfortable with the practice. People may not be appreciative of the concept of their health being in the hands of AI. That being what creates drugs and medicines to potentially try and save their lives. And with the fact that drug and medicine testing is essentially a mystery to the general public, that likely causes even more distrust. If someone already does not trust what is creating medications and is totally unaware of how they are testing them, they’re very likely to completely disregard those announcements as false or impossible to believe. That could even endanger human lives if AI is not as transparent and trustworthy with its machinations and inner workings as possible. Additionally, AI is still incredibly costly for companies and individuals to produce and work with. “Many AI and ML models rely on large volumes of high-quality data to function effectively, but in areas with limited data availability, such as rural or developing regions, these models may underperform” (Alotaibi & Nassif 2024). There are hundreds of areas across the globe that do not have the exorbitant amounts of money that the richest countries possess. AI cost millions of dollars simply to research and develop, much less to implement and use day to day. These resource-scarce areas may not have the necessary budget or supplies to implement AI in their scientific fields. This causes a major disparity in what countries and parts of the world are able to reap the benefits of generative AI and machine learning in their fields. This would cause an even greater divide between richer and poorer countries and their scientific endeavors. Classically, the richest countries often create the most scientific and medical breakthroughs, while the less fortunate countries often are not able to create such prolific and impactful discoveries. Likewise, policies must be enacted to ensure that AI is spread and shared equally with countries around the world. If they don’t, AI could become even more highly privatized than it currently is. Nearly solidifying the fact that it will essentially be only for the rich to use as exorbitant prices, stripping most of the world from its benefits and rewards.

Security Threats Posed by AI in Scientific Data:

    AI is hardly perfect and is subject to mistakes and misinformation. If it is fed something incorrect, it may output a result that is flawed. Though, it may simply make an error and give a false output that is not in line with the data inputted or with the prompt it received. Additionally, some researchers have used generative AI to write their articles and publications, even accidentally leaving in the prompt “regenerate response.” That prompt is used when you wish for the generative AI to rewrite the output it gave you, meaning the researchers did not write their own papers. They also sometimes failed to proofread their papers and had terms or words that were improperly used as the AI looked for synonyms in their place, such as “random value” being replaced by “irregular esteem” (The Economist 2024). AI as well can be hacked and be the victim of cyber-attacks seeking to leak or manipulate the data that it has been fed. While some data will not have severe consequences, large amounts of data are very fragile and will drastically alter the output with even the slightest mistakes. They are confidential much of the time for a reason, to protect the integrity of the data and from misinformation about it from being spread to the public before the real analyses and conclusions can be reached by the scientific community. A hacker is also able to use AI somewhat as a means of hacking. AI, if programmed correctly, is able to inspect code and other programs and reveal holes in their defenses that those with malintent can use to gain an entrance into the program in question. Additionally, if AI itself is manipulated with malicious purposes, it can be used to spread viruses and malware to computers and programs nearly worldwide. “Malware can spread on these LLMs [Large Language Models] by manipulating the AI tools to produce wrong information” (Veerasamy et al 2024). If you feed a generative AI a line of code and ask for it to help fix it, or to make additions, and it has been manipulated by hackers, it may be programmed to place malware into your code or data. Completely corrupting it with next to no way to trace that corruption back to it, save for a few select circumstances. In the capitalistic system much of the world lives under, competition is bred. However, competition is not always positive. The field of AI is just blooming, and the contest to figure out what companies and people will come out on top of the game is already underway. “In the technology marketplace, the race to dominance can force competitors to waive safety concerns in favor of product adoption” (Veerasamy et al 2024). If companies and individuals are ignoring the security issues that AI face simply to make a quick buck, are we really sure this is something that should be privatized in the way it has been? At the very least, if governments had control over AI, the populace would have some forms of pushback. But against companies, there’s next to nothing people affected by them can do that will truly hit them where it hurts.

Future Trends and Challenges:

    While not everyone may agree on where to draw the line with AI in the scientific data analysis field, there still must be a line drawn somewhere. If the use of AI is left unchecked, the possibilities could become out of control very quickly. Either way, AI will continue to be implemented into scientific fields regardless. It has undoubtedly beneficial qualities that can be used to help with further scientific discoveries and increase research into health conditions that were not possible before. AI can work around the clock unlike humans and is able to be less biased than a human judging the data. “Unlike their human counterparts, robots can work 24/7. Well, at least until they need to recharge their batteries” (The Economist 2024). Humans need sleep, break times, and to take care of their personal lives. But robots and AI have none of those requirements. Meaning that they are, theoretically, the perfect workers in many different ways. The fact that they are able to work so efficiently, and for such a long period of time at once, causes many to feel that AI-powered lab robots will help to usher in a golden era of science. On the other hand, AI will not be perfected for quite some time and may never be. But humans might not be too willing to accept AI-powered robots into their workplaces, especially specific groups of people. AI has been unfairly distributed among groups of scientists. Though, this is partly due to systemic issues, and peoples’ reactions to them. “…the benefits of AI aren’t evenly distributed: they are lower in disciplines with a higher share of female and minority researchers” (Gao & Dashun 2024). Historically, women and minority workers and employees have been treated as lesser when compared to their white, cisgender, heterosexual, male coworkers and colleagues. This is not to say that those men are unqualified or undeserving of the same opportunities, but they are often given more than they’re worth. Women and minority workers have always experienced higher likelihoods of being fired or let go, suffering workplace harassment, and not being taken seriously when they speak up about their suffering. Naturally, it would make sense for those same people to be untrustworthy of AI, a robot that can possibly do their exact same jobs except require no time off, no pay, no breaks, and absolutely zero benefits. If you knew that you were more likely to be let go over the male coworker described above, would you really be jumping at the opportunity to let a robot take over part of your job? AI and will continue to be used more and more as time progresses, but the setbacks and deficits that AI currently experiences have to be corrected before it can truly be applicable nearly everywhere.

Conclusion:

    AI can be both beneficial and detrimental to the fields of scientific data analysis. It is an imperfect tool that has yet to be truly refined into a polished feature. Its current uses are already many and great, but it could become so much more if there was a larger push to secure the data used by AI. However, that does not mean that all human analysts should be replaced by machines. AI will always make mistakes that the human brain is able to figure out with our unique processing patterns. Having human researchers working in tandem with machines and each other will create a truly remarkable environment for scientific research and discovery, one that will foster new revelations and creations much faster than ever before. AI should be used in the analysis of scientific data but should never replace human workers and scientists. It has already begun to bring about extreme change in the scientific world just within the last few years, and it will continue to do so much more if we enact policies and regulations to contain AI’s reach and keep it from overstepping into roles and fields that require a human touch. There is no perfect answer, and it is something we will need to discuss for years to come, but AI will continue to be used, and its uses will expand further and further. At the end of the day, AI will not be disappearing anytime soon, that much will always be certain. AI is one of the most powerful tools a researcher or scientist can use in this modern day and is something that we should all be trying to implement in positive, constructive ways in a plethora of fields. Of course, AI may not have the required metaphorical touch for certain fields that deal with very human subjects like mental illness, sexuality and gender identity, and art to name some, but that does not mean it is useless. In fields that require an unflinching, objective truth like science and math, AI will excel. We simply have to be certain that we control it, lest it run rampant beyond human grasp.