AI in science isn’t good or evil. It’s just a tool.

Stanley Kubrick’s film 2001: A Space Odyssey opens with the words “The Dawn of Man.” (Spoiler alert if you haven’t seen it.) In the scene, a group of early humans struggles to survive in the African desert—scavenging food, fending off predators, and generally getting the short end of the evolutionary stick.

Then, one of them picks up a femur bone and discovers it’s great for things like smashing the skulls of proto-pigs. Suddenly, the tribe has food, safety, and a competitive edge. Moments later that same bone becomes a lethal weapon against a rival tribe.

Kubrick’s lesson: any tool powerful enough to help can also be used to harm.

Artificial intelligence is a lot like that bone. Not inherently good or evil—just very useful and powerful. And like any tool, how it’s used matters far more than what it is.

Sensationalism Sells, but Reality is More Complicated

Most headlines portray AI as either humanity’s salvation or its doom. It’s either the most important invention since the microprocessor or the end of civilization as we know it.

The truth, of course, is far less dramatic. AI is simply a very competent system for performing certain types of tasks, especially those involving scanning and structuring large volumes of distributed data. That’s what it’s good at.

What it’s not good at is judging the quality of that data, spotting flaws in its own logic, or determining when it’s strayed too far from the truth. We all know that AI can “hallucinate”, generate convincing but entirely false conclusions, and do so at scale, making the errors harder to detect.

None of this means AI is bad. It means AI is a tool; a tool that requires a competent operator to handle it.

The Real Question Isn’t “Is It True?”; It’s “Is It Trustworthy?”

AI can parse information at a speed and scale no human can match. But if we have to manually double-check every result, especially in something as complex as biomedical research, then what’s the point of using it?

So, when we’re using it to make critical decisions, how do we ensure AI’s output is trustworthy?

Do we use more AI to check the work of other AIs? Do we build non-AI systems to validate results? Do we rely on human experts to analyze every detail?

The answer is: all of the above.

Meet Tessa, the Trust Framework

At Siensmetrica, we’ve been developing a framework for trustworthiness we call “Tessa”.

Tessa is an anthropomorphized term for “TES”, short for Transparency, Explainability, and Significance. It is a framework designed to evaluate the underlying components of the research; not just what it says, but where the source materials came from, why it says what it does, and whether the conclusions it reaches are meaningful.

Unlike large language models that predict what comes next, Tessa performs a molecular-level inspection of data inputs. It looks at each data point used to generate the results and asks:

• Is the data transparent? Can we see what was used?
• Is the reasoning explainable? Do the authors show how they arrived at this conclusion?
• Is the data significant? Was it appropriate to include it in the analysis, and are the results useful?

Currently, Tessa’s framework analyzes 163 (and counting) distinct data vectors. The result isn’t a binary “right” or “wrong” judgment but a trust score from 0 to 100. The higher the score, the more confidence users can have in the data’s trustworthiness.

AI Can’t Replace Human Intelligence, It Requires It

When Tessa scores a piece of research highly, it doesn’t mean the answer is correct. It means the process behind the answer is strong and reliable, i.e., trustworthy. It gives researchers and practitioners a quick way to separate high-quality information from low-quality noise and focus their time where it matters most. But while AI can process oceans of information in seconds, it can’t replicate human judgment. And when it comes to science, medicine, and the decisions and policies made from it, judgment is everything.

The Bottom Line

AI isn’t inherently good or bad. Like that femur bone in 2001, it’s a tool and it can be used to help surface truth or harm by reinforcing falsehoods. The only way to separate one from the other is to build systems like Tessa that help us analyze the reliability of information and give us a metric for how trustworthy it is.

Because in a world where the available research is doubling every 10 weeks, identifying trustworthy information, not creating quick summaries, is the real breakthrough.

The post Is Artificial Intelligence Good or Evil? first appeared on Tessa AI.

Defining trust

Before I launch into how to restore trust, I should clarify what I am talking about when I use the word. Trust, simply put, is confidence in or reliance on the ability, faithfulness, or accuracy of someone or something.

While trust is often found in the company of “truth”, the two are not the same. “Truth” is a result, whereas “trust” is focused on the means and methods used to arrive at a result. It is an important distinction because in the world of scientific research we don’t always discover a truth, but we should always be able to determine the level of faithfulness and accuracy in the means and methods used to perform that research.

In other words, it needs to be trustworthy.

Components of Trust

Trust is a sort of cognitive shortcut that allows us to rely on something without a deep analysis of that thing. If it meets certain criteria, it can be trusted. Psychologists have studied the concept extensively and identified a number of cognitive and social factors that influence how people perceive someone or something as trustworthy (and vice versa).

Even if you have not studied the concept, you are still probably familiar with many of these factors (aka “heuristics”): source credibility or authority, bandwagon effect, perceived motive and intent, consistency and coherence, willingness to engage and revise, confirmation bias, etc. In all, there are around a dozen of these elements; too many for the purposes of this article.

Instead, I have borrowed a simplified framework for trust from computer science and neural networking where they have taken the various components of trust and put them into three interrelated conceptual buckets: Transparency, Explainability, and Significance (TES).

• Transparency can be summed up as “show your work.” In scientific research it is the clarity and fidelity of the methodologies and data sources, affiliations of the researchers, funding sources, a level of detail that allows others to repeat the research described, and so on.
• Explainability is the consistency and coherence of the information and the ability to support it if it contradicts other information. It also factors in elements like the level of uncertainty, the ability to address questions and criticisms, and details that help shed light on how the conclusions were reached.
• Significance is a measure of the relative importance of both the components and results of a piece of research. Components include the type and nature of study, the reputation of the authors, institutions and publications, etc. It also factors in peer review commentary, recency, citations, and other elements that help determine the overall impact of the information.

Each component of TES has many interrelated sub-components so the weighting of each element varies based on the information being analyzed. However, placed on a scale of 0 to 100, the closer each component is to 100, the more “trustworthy” it is (though scientific skepticism makes it extremely unlikely that anything or anyone could ever have a perfect 100 across all three categories).

Restoring Trust in Science

In my previous post I identified a number of areas where the scientific research and medical communities had significant failures in each area of TES, leading to an erosion of public trust in science. Whether it was a failure to disclose financial incentives (transparency), the repeated use of fraudulent data (explainability), or the resistance of highly regarded scientists in acknowledging their inaccuracies or errors (significance), or all of them collectively, the damage done has been significant and as such, will take some time to repair.

The good news is the scientific community is more aware of these problems than ever, and many groups are now working on a variety of solutions, both traditional and technological. The quickest fixes lie in a return to the scientific principles of transparency (especially funding sources), open analysis, and a willingness to address conflicting or contrary information in a timely manner.

Longer term the solutions will most likely be technological. Open access and pre-publishing platforms have turned the research spigot wide open, doubling the amount of available research in medicine alone every two months. Parsing and processing that volume of information will require methods faster and more robust than the traditional, more measured and time-consuming methods of peer review and journal publication.

Not surprisingly, AI is playing a large role in attempting to get more control over the research deluge with numerous tech firms devising innovative and useful ways to search, process and digest information. Other organizations like RetractionWatch are doing important work in surfacing corrections and retractions in the pursuit of scientific progress.

For our part, here at Siensmetrica, we are focused on applying deep data science and multi-vector analysis to identify and score research based on TES principles. Our aim is not to supplant the rigors of the scientific method, but to provide scientists and medical professionals with an objective way to assess the trustworthiness of a piece of research—a tool for guiding further exploration. The goal is not to always be right, but to be transparent and consistent so that we can restore belief in a system which, though not always perfect, can be relied upon to be heading in the right direction.

If we are successful, “trust the science” will be the gold standard once again.

The post How to Restore Trust in Medical Research first appeared on Tessa AI.

For decades, the phrase “follow the science” conveyed a sense of unwavering trust. It implied confidence in a system dedicated to advancing human health through rigorous inquiry. However, recent events have revealed significant fractures in this trust, exposing systemic issues that threaten both public confidence and the very progress of scientific discovery.

The scandal surrounding Alzheimer’s research serves as a stark example. As reported in The Chronicle of Higher Education, a substantial portion of foundational research over the past two decades was based on manipulated or fraudulent data. This deception led to the misallocation of tens of millions of research dollars, wasted the efforts of numerous scientists, and, most critically, delayed the pursuit of effective treatments and a cure. The implications are profound: how many lives could have been improved, or even saved, had research not been misdirected by fraudulent findings?

This crisis of trust extends beyond Alzheimer’s research. The decline in public confidence in medical institutions and scientific integrity is widespread. During his recent Senate confirmation hearing for FDA commissioner, Dr. Marty Makary of Johns Hopkins University highlighted this alarming trend:

“Trust in doctors and hospitals is currently in a critical state. It was 71% in 2019… now it’s down to 40%. That’s a 31-point drop in my profession. This is a significant problem. If we develop a cure for pancreatic cancer but only 40% of the public is willing to accept it due to distrust, the medication’s effectiveness is essentially reduced to 40%.”

Dr. Makary’s statement underscores the real-world consequences of a public that questions the integrity of medical research. If individuals decline treatment or doubt the validity of scientific findings, even groundbreaking discoveries lose their potential impact.

A central factor in this crisis is the influence of financial incentives in research. Whether it’s the pursuit of lucrative grants, the pressure to publish impactful findings, or the financial interests of pharmaceutical companies, the incentives for scientists are not always aligned with the pursuit of objective truth. This misalignment has contributed to instances of manipulated studies, suppressed data, and an overall erosion of faith in the scientific establishment.

It is essential to distinguish between scientific truth and public trust. Science is an iterative process, and current understandings may evolve as new evidence emerges. This is inherent to scientific progress. However, public trust should remain steadfast. The expectation is not that scientists will always be correct, but that they will be transparent and honest in their pursuit of knowledge. When financial incentives, personal ambitions, or institutional pressures supersede the unbiased search for truth, trust inevitably deteriorates.

This decline in trust does not imply that science is irreparable, nor does it suggest that all research is flawed. It does, however, indicate a deep-seated problem that requires attention. Fortunately, awareness of these shortcomings is growing within the scientific community, and efforts to address them are underway.

In the second part of this series, I will explore potential solutions to rebuild trust in medical research—how transparency, accountability, and systemic reforms can restore confidence in a field that, at its best, has the power to transform lives. For now, the reality remains: trust in medical research has been compromised, and without decisive action, the consequences will be felt for generations.

The post The Erosion of Trust in Medical Research: A Crisis Decades in the Making first appeared on Tessa AI.

Open access and preprint articles are overwhelming the industry with research; we need a more trustworthy, explainable and significant way to separate the high quality data from low quality data

I have been in the healthcare field for a couple of decades now; long enough to see a lot of changes. Faxes replaced by email, paper replaced by electronic healthcare records, office consults replaced by smartphone apps. But nowhere has there been as much change as in medical and scientific research publishing.

As recently as 2009 research publishing was mostly the purview of a relatively small group of academic publishers controlling the majority of journals. As gatekeepers of information, the lifeblood of science, these journals were able to generate substantial revenues through both fees to subscribe to more prestigious publications, and fees to publish (known as author publishing charges or APCs) in lower impact partner publications.

The Rise of Open Access

Things began to change in 2004. The U.S. National Institutes of Health (NIH) Public Access Policy mandated open access to research they funded, and Europe’s Plan S soon followed suit. Digital publishing and Internet access made widespread electronic distribution possible; open access and preprint publishing platforms soon followed, opening the research access floodgates.

In The Changing Medical Publishing Industry: Economics, Expansion, and Equity, a paper published by a group of physicians in the Journal of General Internal Medicine (JGIM) in July of 2023, the authors note the number of open access journals over the past 20 years has risen from near zero to over 16,000.

Source: The Changing Medical Publishing Industry: Economics, Expansion, and Equity, Springer.com

The authors note that we are only beginning to understand the opportunities, and consequences, today.

…open access will result in wider dissemination, and preprint platforms with faster communication of science… However, consumers of research may have more difficulty assessing the validity of conclusions from the increased number of research reports… Rapid expansion of published research brings with it the risks of prioritizing quantity over quality of medical research. This poses risks to patients if clinical decisions are influenced by shoddy science.

Identifying Quality Among Quantity

Of course, quantity over quality has always been an issue in scientific research, which is why there are checks like conflict disclosures, statements of limitations, the h-index, and peer review. But even with these, sheer volume and velocity of newly published research, and the opaque nature of peer review makes the task of identifying valuable knowledge in a rising sea of information extremely difficult. The JGIM paper’s authors note that current methods are better than nothing, but still a far from a complete solution:

“…peer review remains an important part of the scientific method. We acknowledge that the empiric benefits of peer review remain debatable, and it may not filter high-quality from low-quality research as well as we like to believe. However, we believe it is better than an alternative system in which there is no formal process for peer input and constructive critique.”

The writers go on to suggest that newer open models including engagement via comments on preprint platforms and community peer review via systems like PubPeer have potential. However, my own experience with healthcare providers, researchers and other consumers of this information, indicates that inputs like these will probably fall short. They’re labor-intensive and, more importantly, as we’ve seen in other open discussion platforms, subject to manipulation by bad actors and robots. As a result, for the Open Access research model to succeed, new models for vetting and valuing research will also be required.

Developing New Metrics for Research Quality

One of the biggest problems with open access and preprint literature is also one of the oldest: “information overload.” Researchers at Oxford University define this as:

…the situation that arises when there is so much relevant and potentially useful information available that it becomes a hindrance rather than a help.

They note that this overload creates a condition creatively called data smog, which is a cloudy, unhealthy, perception of what information is actually available and whether it is of value. Anyone who has ever done a web search on a topic that seems fairly specific and straightforward only to be presented with a million results knows the data smog effect and the skepticism in any results it leads to.

Commercial search engines like Google have developed sophisticated models that do a pretty good job at identifying and presenting relevant information in search. Unfortunately, people have developed techniques in “search engine optimization” (SEO), to game search results in such a way that unless you are a sophisticated search engine user familiar with the commands and parameters for creating highly specific search queries, you are still likely to be showered with mostly irrelevant information.

The question is then, if a near-trillion-dollar search engine giant can’t create a reliable method for identifying and presenting high quality, trustworthy information, how is a frontier community like open access academic publishing supposed to?

Trust, Explainability, and Significance

Fortunately, we have examples of how this might be accomplished from applications in other scientific fields. Researchers are leveraging advances in data science, neural networking, machine learning, and to a lesser degree, artificial intelligence to successfully identify solutions for extremely complex problems. Rather than relying on Boolean parameters like AND, IF and NOT, they are using multi-modal analysis of vast quantities of information to develop models that identify and present the most promising data. Computer scientists working in these areas often refer to three very human terms—Trust, Explainability, and Significance (TES)—as the underpinnings of how they assist human beings in identifying high quality information in an otherwise limitless ocean of data.

TES is already being used to explore new avenues of research and help make incredible advances in a range of highly complex fields from molecular chemistry to financial market behavior analysis. Significantly, these fields have an open access history that is more agile and mature, but no less complex, than that of medical research publishing. As a result, leveraging TES-based technologies for both the discovery of high-quality information, and assisting human beings in confident decision-making based on it, offers considerable promise in healthcare research and clinical decision-making.

Traditional methods for establishing the validity and value of research will still have value in an open access, open distribution publishing model. But, without some means of separating the high quality from the low in an industry that is doubling the amount of available information every three months, information overload will grow, and confidence will fall. Incorporating Trust, Explainability, and Significance concepts into medical research exploration and search technologies offers a way to confidently advance and accelerate medical knowledge and improve healthcare delivery and outcomes.

The post Medical Research Publishing Needs New Metrics first appeared on Tessa AI.