Image credit: Cornell
You are late for a work presentation due to rush-hour traffic, and you think to yourself, “if only the car could do this highly repetitive action autonomously and I could present at my meeting from the road”. But you have to feed the gas pedal every five seconds to move less than a few yards. If Silicon Valley’s stalwarts such as Waymo, Uber, and Lyft had delivered on their decade-old promises, this could have easily been a commonplace sight by now where cars would drive us to and from destinations without any human intervention.
In March 2021, Lyft relinquished its hope to create a fully-autonomous self-driving system and offloaded its self-driving division to Toyota’s Woven Planet Holdings subsidiary for $550 million. Rethink Robotics, a venture led by AI stalwart Rodney Brooks, also closed its doors in 2018 after dedicating a decade to ushering collaborative efforts between humans and intelligent robots for industrial automation. Additional fiascos in the robotics & AI space gave the impression that all wasn’t well in the much-hyped “AI kingdom”.
Although many quickly jumped the gun and declared that AI was doomed for another winter, people seemed to have missed the fact that these incidents were merely a surface-level effect of a much deep-rooted problem. For instance, Rethink Robotics’ predicament was a direct consequence of its obsession with creating the smartest robot, even if that was not the need of the hour. As a result, Rethink Robotics ended up creating overly complicated systems that were economically infeasible to scale, so they were futile to customers. Similarly, leading up to its divestment in Aurora, Uber ATG was plagued by a myriad of incidents ranging from the fatal pedestrian incident in Phoenix in 2019 to the infamous lawsuit against Anthony Levandowski.
If AI was not the reason for these grandiose failures, then why was it thrown under the bus? It’s mainly due to the fixation people have developed with the field because of the hype around it. To understand that, let’s look at how AI gained its status-quo.
When John McCarthy coined the term “artificial intelligence” in 1956, it was only used in cliques of renowned researchers such as him and Marvin Minsky. However, due to certain technical breakthroughs in AI research and several entertaining but over-zealous interpretations by the movie industry, the term trickled into the general vocabulary. The glaring disparity in domain knowledge between the general public and AI scientists resulted in the two communities forming starkly different notions about AI.
The business world also left no stone unturned when it came to fueling and riding the AI wave. As a result, a robot on a death killing spree in Terminator is considered AI. Jarvis—Iron Man’s brainchild—acts as a sentient powerhouse running everything from his doorbell to his kick-ass suits is also considered AI. A “smart” workout machine that merely adjusts resistance automatically is also labeled as “AI-powered”. The next few sections highlight how the mainstream media and desperate businesses are behind the hype around AI, which is blurring the line between actual AI and ordinary computer applications.
Regardless which side of the argument one falls on about the AI hype, it is undeniable that the AI hype is pervasive. The ballyhoo that originated in the mid-20th century has not only affected the education and business sectors but has also made a lasting impact on our perception about AI. So, who should receive the credit for this undying hype that has not only stood the test of time but has also altered the outlook of our species as a whole? It certainly cannot be one entity or segment. The AI hype is analogous to a virus that finds a new “host” every so often, and that has kept it from dying. Initially, it was the scientific community and the research labs that hosted this virus. Next, it was the defense industry and Hollywood. Lately, businesses and the general public have joined in to keep the virus afloat. Let’s take a quick look at how the hype has flourished and propagated along all these years.
AI hype is analogous to a virus that finds a new host ever so often, and that has kept it from dying.
Unrestrained excitement is inevitable whenever there is a breakthrough in a field—especially in tech. The early fervor around AI originated in 1958 when Frank Rosenblatt introduced perceptrons: a form of neural network with adjustable weights. Rosenblatt’s invention emerged out of his passion for neuroscience. Along with this groundbreaking invention, Frank Rosenblatt also put forth a bold prescience, “perceptron may eventually be able to learn, make decisions, and translate languages.” Although he and many other researchers were indeed correct, the proposed timelines regarding the capabilities of AI were quite quixotic.
Rosenblatt sent the world into a frenzy when he unveiled a bespoke machine with motorized potentiometers that not only ran perceptrons faster than any other computers of his time and but also classified different images of shapes or letters using learning algorithms. Such demonstrations and advancements had people convinced that AI was going to change the world within the upcoming decade. This impetus resulted in huge investments from both the Defense sector and the private sector.
The Navy revealed the embryo of an electronic computer that it expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence.” - The New York Times, 1957
Rosenblatt, along with other researchers, was convinced that he could replicate the function of neurons in the brain and create a computer that could mimic a human’s intelligence, hence the name artificial intelligence. This fixation with mimicking human intelligence was what started the AI hype and has kept it alive till date. Ironically, it would also end up being the reason for one of the major dry spells in AI development: the AI winter.
The initial claim by AI researchers that the learning and functioning abilities of AI algorithms would be commensurate with those of a human brain provided such a huge impetus that it still shapes the sentiment of the general public to date. Consequently, Hollywood—like any other business — also decided to jump onto the AI wagon. For instance, Hollywood fully exercised the liberties of the sci-fi genre to give Terminator Mr. Olympia’s arms with an AI brain (without any explanation about why it had a thick Austrian accent). Thus, when the concept of AI having sentient capabilities was reinforced by hallmark movies and shows such as The Matrix, I, Robot, and Westworld, viewers developed the notion that AI could do all physical tasks like a human with 100 times the strength and efficiency. A recent article from Skynet Today succinctly explains how the Hollywood definition of AI reinforces an already common myth that present day AI has some amount of human-like agency, when in fact this is not really the case.
I remember pitching an idea for a smart workout bike that only proposed the use of AI to gather user inputs and learn about their workout habits, but the actual product had no AI integration whatsoever. After my first pitch, one of the mentors mentioned that I should talk more about the “AI aspect” of the product to catch the VC’s attention. It was during this market research for my fitness bike that I came across numerous smart and “AI-powered” workout platforms that were merely adjusting resistance of a pulley electronically but were still funded extensively. In an op ed by Kathy Pretz published in IEEE Spectrum in March 2021, Michael I. Jordan, a leading AI researcher at the UC Berkeley, mentions that many underlying systems of such products do not involve high-level reasoning or thought. The systems do not form the kinds of semantic representations and inferences that humans are capable of. They do not formulate and pursue long-term goals.
Virtual Reality (VR) is a great technological example of how venture capitalist (VC) firms can sometimes make a promising but underdeveloped idea seem ready to dominate the world. VR startups and businesses captured the imagination of the world for most of the last decade. Any idea even remotely related to VR would promptly create an exorbitant amount of excitement among VC firms and soon receive their monetary blessings as well. The firms didn’t seem to care too much about the scalability or even the feasibility of the proposed idea as long as it mentioned the term “Virtual Reality”. AI businesses are enjoying a similar high right now, where all they must do is pad their business pitches with AI-related terms such as Machine Learning, Deep Learning, Neural Networks to get the VCs jumping in their seats.
It would be remiss to state that AI hasn’t impacted the lives of billions of people over the last several decades. However, AI has been haunted by the ghost of its own hype ever since its inception. As a result, the reality of AI has never seemed to align with the expectations of people. Moreover, the AI startups that didn’t make it, AI wasn’t the primary reason for their failures–it was the obsession to make AI something it wasn’t.
This obsession is a direct repercussion of the fairytales knitted around AI by different communities over the last few decades. It all started with the nonviable promises made during the nascent stages of AI development. The hype since then has been constantly fueled by the various over-the-top interpretations shown by Hollywood flicks and TV shows. The general public has always seen advancements in AI as mere stepping stones to a robot with superhuman intelligence. This somewhat flawed notion is mainly a consequence of Hollywood’s obsession with creating “AI-based sentient beings”.
More recently, the VC firms have arrived onto the scene to invest exorbitant funds in the AI field. And to recover their investments, these firms have used all the tools at their disposal, even if that means tampering with the public’s perception of AI and generating baseless hype around mediocre products.
As AI becomes more pervasive, there are several ways the uninitiated can avoid misinformation about the topic. One important step would be to use reliable sources for forming opinions, such as podcasts from AI researchers directly. By getting information directly from the horses’ mouth, one can avoid falling for marketing gimmicks. It would also be helpful to question AI marketing campaigns. If a product claims to be AI-enabled, investigate if it actually is making its own decisions or is it using some preloaded conditional statements that only work for pre-determined scenarios. In the end, it’s important for everyday users to judge AI, like any other technology, based on the functional efficacy of the currently implemented solutions rather than what emanates from hype campaigns.