Hello World

Introduction: How Algorithms Shape Our World

Physical objects often carry the hidden biases of their creators. Hannah Fry points to the low bridges leading to Jones Beach, designed by Robert Moses to be too low for buses. This intentional design choice effectively barred poorer citizens who relied on public transit from accessing the park. Similar invisible barriers exist today within technology. From soap dispensers that fail to recognize dark skin to internet filters that block entire towns due to their names, these tools reflect the oversights or intentions of the people who built them.

Algorithms now act as the digital infrastructure of our lives, influencing everything from medical diagnoses to court sentences. These mathematical rules are not inherently good or bad; their impact depends entirely on how humans use them. By examining these systems, we can better understand the relationship between people and machines. Ultimately, the goal is to recognize when to trust a machine's logic and when to rely on human judgment, ensuring that technology serves to improve society rather than limit it.

In 1997, the world witnessed a historic clash between human intuition and machine calculation when chess grandmaster Garry Kasparov faced IBM’s Deep Blue. While Kasparov was famous for using psychological tactics, the machine was immune to such pressure. However, the engineers behind Deep Blue understood something critical: to beat a human, the machine had to exploit human fallibility. They programmed the computer to occasionally pause for several minutes before moving, even if it had finished its calculation instantly. This thinking time led Kasparov to believe the machine was struggling with the complexity of the game. When Deep Blue later made a move that seemed uncharacteristically human by anticipating a trap, Kasparov became unnerved, eventually losing his confidence and the match. Kasparov’s defeat was not just about the machine's processing power; it was about his own misunderstanding of what the algorithm was actually doing.

At its simplest level, an algorithm is just a step-by-step procedure for solving a problem, much like a cake recipe or IKEA assembly instructions. In the digital world, these instructions are translated into mathematical code to perform four primary tasks: prioritization, classification, association, and filtering. These systems generally follow two paths. Rule-based algorithms use direct instructions written by humans, making them easy to audit but limited by the programmer's knowledge. Machine-learning algorithms, a subset of artificial intelligence, learn by trial and error from vast amounts of data. While these are incredibly powerful at tasks like speech recognition, they can be mysterious; a machine-learning tool might correctly identify a car but change its mind to a dog if a single pixel on the wheel is altered.

Our relationship with these tools is often defined by a dangerous mix of over-reliance and sudden distrust. This blind faith was evident when a driver named Robert Jones nearly drove his BMW off a hundred-foot cliff because his GPS insisted a narrow dirt path was a major road. Even as the road vanished, he trusted the machine over his own eyes. Research shows this bias is widespread; in one study, undecided voters were easily swayed toward a political candidate simply because a rigged search engine placed that candidate’s name at the top of the results. Because these systems are right most of the time, we stop questioning them, often unaware that we are being manipulated or that the authority of the machine might be masking human error.

This misplaced trust becomes life-altering when algorithms are used in public policy without transparency. In Idaho, a budget tool used to calculate Medicaid benefits suddenly slashed support for thousands of disabled residents. For years, officials defended the cuts as the result of a proprietary trade secret. When a lawsuit finally forced the software into the light, it was revealed to be nothing more than a buggy Excel spreadsheet filled with statistical flaws. The objective machine was actually just a collection of human mistakes wrapped in code. This highlights a recurring theme: we often attribute superior judgment to a computer simply because it produces a number, failing to realize that many algorithms are just natural stupidity disguised as artificial intelligence.

Finding the right balance of power between humans and machines remains a challenge. History shows that human intervention can save lives, as it did in 1983 when Soviet officer Stanislav Petrov correctly guessed that a computer's nuclear launch alert was a false alarm. Conversely, humans can cause disasters by over-ruling a perfectly functioning safety system, as seen in a tragic 2015 UK rollercoaster crash where engineers bypassed an automated emergency stop. While mathematical models almost always outperform human intuition in technical predictions, we suffer from algorithm aversion, meaning we are far less forgiving of a machine's mistake than our own. To navigate this age, we must learn to treat algorithms neither as gods nor as junk, but as tools that require constant, critical human oversight.