Algorithmic Information Access is the New Smoking

Algorithmic Information Access is the New Smoking

Chirag Shah, University of Washington

Smoking used to be cool. Yes, it’s still considered “cool” by some people, at some times, and at some places. But in general, at large, smoking has been on decline according to the American Lung Association. Why am I talking about smoking? That’s because there are things we could learn from how smoking prevailed at some time in our history, how we recognized its harm, and how we came to reduce it as a society. I’m talking about smoking because our addiction to online information access is similar to what smoking was several decades ago, and if we were to address this addiction, we could find possible solutions from how tobacco users, media, and government played a role in fighting against the tobacco industry.

—We are addicted to online information access, devices, and services.—

Let’s start with the problem. Almost all of the information we access these days — whether through searching, browsing, or from recommendations (think about news feeds), we get it through some algorithm behind the scene. Sure, we are still using bookmarks and there are times we just know where we want to go (e.g., a travel site) and either we can type in a URL or search for a name and immediately click on the first result. But think about all other times when we need a system’s support — running keyword-based queries, scrolling through a feed on social media, and taking recommendations from our favorite movie or music streaming service. We rely on those algorithms to access information.

Is this really bad? No, at least not on the surface. After all, we built these systems to solve a problem — a problem of relevance. When there is too much information and we don’t have enough capacity to sift through it as we look for what may be relevant to us in a given need, we could use a system’s support to do just that. After all, computational systems have been used for searching, sorting, and retrieving items much before the advent of the web. We have understood these systems for a long time. Well, that was the original intent — matching an information need based on relevance. But today’s information access systems do more than running those search and sort algorithms that a freshman computer science major learns about.

These algorithms need to serve their “masters.” While the information seekers are the users, the masters are those baking these algorithms in their systems and services that they run commercially. How do they make money? I won’t bore you with the details as I have written about it elsewhere, but just note that user engagement is what pays for the seemingly “free” services for information access. The more users use their service, the more often they use it, and the longer they stick around, the more revenue these services can generate. So what do you do to keep people coming back for more? Or in other words, make your services addictive? You give them more than just what’s relevant. You give them sensational stuff, you give them entertaining and self-confirming information, you give them content that trigger strong reactions like hatred and fear. And as we click, scroll, and consume more of these, the algorithms learn that they made the right decision recommending that content. They, in turn, bring more of such content to us, and the vicious cycle continues.

How do we break this cycle? This is where I believe we can look at what happened with tobacco consumption. There was a time when not only we didn’t know enough about the harms of tobacco, but smoking was considered healthy. Obviously, we had no reason to quit smoking, and the tobacco industry had no reason to stop selling their products. But as we understood more about the harms that tobacco causes to health, we needed a change. Easier said than done, because who will break the cycle? Tobacco users and tobacco producers were locked in a very tight relationship of dependency. Addiction, individual and societal-level, is also not easy to just break. And until early 1950s, we didn’t think it was necessary. But the new realization about serious harms to our health from tobacco uses started stirring some discussions and potential remedies. We needed to and wanted to curb the use of tobacco. But how to do it when it was integrated so tightly in the fabric of our society? The solution came through multiple avenues and tactics.

I believe there are many parallels we can find from tobacco addiction to information access addiction. As a first point, most of us with mobile devices and internet connection regularly want and need access to online information. On the other side, there are plenty of sites, services, and sources that want to provide, facilitate, or mediate that. In other words, there is lots of demand and lots of supply; there is a market. And when there is a market, there is profit to be made in a capitalist economy. Let’s admit it — these online services are not on philanthropic missions. They are businesses with investors and ambitions. Many of them may have started with a vision to change the world for the better, but almost all of them have transformed from that garage or dorm initiated startup to multinational companies with quarterly business objectives and long-term growth plans.

We the users, on the other hand, have offloaded a lot of our decision-making and agency to these services. At first, this was necessary, then convenient, and now, it’s an addiction. Don’t get me wrong, the necessity and convenience are still important, but addiction (can’t put our phones down) is an unwelcome side-effect.

The companies providing access to information (e.g., search engines, social media, e-commerce sites) want us to keep returning to them, using their services more frequently, and spending more time with them. That allows them to sell more ads and products, collect more data, and tighten their grips on us. The users like the value they derive from the continual relationship with these services. Everyone wins, right? Perhaps not. There are harms to individuals and society due to this addiction. It is no secrete that many of us are victims of doomscrolling, taking a huge toll on our mental health. Or that our attachment to online sensationalism leads to misinformation that impact the whole society.

First, we did user awareness and education — putting labels on the box of cigarettes and running ad campaigns on TV and radio. Sure, one could still buy the cigarettes, but now they are increasingly becoming aware that it could kill them.

Second, we made it more restricted and costlier to buy tobacco products. Instead of having these products available everywhere, they were now getting restricted to certain kinds of stores and counters. You needed to be of certain age. And perhaps more importantly, it started costing a lot more due to heavy taxation on it. Now, one would really think twice before buying an expensive pack of cigarette for which they had to go out of their way to purchase.

Third, we integrated considerations for smoking in other parts of our lives. On a healthcare insurance, it became a mandatory field to ask if you smoke. The same goes for life insurance — if you smoke, you had to pay more, and in some cases, may not even be able to secure that insurance.

Fourth, we invested in alternatives to smoking to make it easier for those wanting to quit. Nicotine patches and nicotine gums are some examples. There are clinics, rehabs, and many other options available to those who need help to end their addictions.

Fifth, we went after the big tobacco. They could no longer suppress research on links between tobacco and cancer and other health-related harms. They could not sell their products to minors, or get away with creative marketing and products such as e-cigarettes.

Sixth, and perhaps the most important one — we made smoking socially unacceptable. You couldn’t smoke on airplanes or buses. Then we started banning it in hotels, restaurant and many public places. Sure, some places still provide ‘smoke-friendly’ spaces, but they are not guaranteed, often hard to find, and for those on the fence, not worth the effort. It’s no longer “cool” or socially acceptable for one to smoke in most places and most situations.

Of course, there are also differences between tobacco addiction and digital addiction. It is not because of their similarities in the nature of the problem that I am comparing them here; but rather it is in their similarities of the solutions. Our addiction to tobacco didn’t change overnight. It took efforts from multiple sides over many decades to do that. Similarly, we are not going to simply ask Amazon, Google, or Facebook to fix something or for us to quit using some of these services. In fact, unlike tobacco, there are many good uses of these online services with the right value function (more on that some other time). We just need to be able to break the vicious cycle of (1) what an average user finds engaging; (2) how a typical information access system optimizes for user engagement; and (3) how much a user trusts rankings and recommendations of these systems lead to misuse of these users’ trust and mischaracterization of a system’s trustworthiness.

Just like the case of the tobacco, the first step is for us to recognize the danger of algorithmic information access. No, I’m not trying to be paranoid here or suggest that shopping on your favorite e-commerce site can kill you. But we need to recognize that these products are not there to just serve us; they are addictive and profit driven. Once we recognize this, we can start focusing on constructing a healthier relationship. And for that, we can take some hints from how tobacco consumption behavior changed over a long time. In other words, there is no quick fix here and we need actions from multiple stakeholders — from users, from developers, and from policymakers.

Cite this article in APA as: Shah, C. (2021, December 27). Algorithmic information access is the new smoking. Information Matters.  Vol.1, Issue 12. https://r7q.22f.myftpupload.com/2021/12/algorithmic-information-access-is-the-new-smoking/

Author

  • Dr. Chirag Shah is a Professor in Information School, an Adjunct Professor in Paul G. Allen School of Computer Science & Engineering, and an Adjunct Professor in Human Centered Design & Engineering (HCDE) at University of Washington (UW). He is the Founding Director of InfoSeeking Lab and the Founding Co-Director of RAISE, a Center for Responsible AI. He is also the Founding Editor-in-Chief of Information Matters. His research revolves around intelligent systems. On one hand, he is trying to make search and recommendation systems smart, proactive, and integrated. On the other hand, he is investigating how such systems can be made fair, transparent, and ethical. The former area is Search/Recommendation and the latter falls under Responsible AI. They both create interesting synergy, resulting in Human-Centered ML/AI.

Chirag Shah

Dr. Chirag Shah is a Professor in Information School, an Adjunct Professor in Paul G. Allen School of Computer Science & Engineering, and an Adjunct Professor in Human Centered Design & Engineering (HCDE) at University of Washington (UW). He is the Founding Director of InfoSeeking Lab and the Founding Co-Director of RAISE, a Center for Responsible AI. He is also the Founding Editor-in-Chief of Information Matters. His research revolves around intelligent systems. On one hand, he is trying to make search and recommendation systems smart, proactive, and integrated. On the other hand, he is investigating how such systems can be made fair, transparent, and ethical. The former area is Search/Recommendation and the latter falls under Responsible AI. They both create interesting synergy, resulting in Human-Centered ML/AI.