AI Cleansing: The great recent untruth of Silicon Valley
“Could you revisit all the previous presentation slides and substitute the term ‘cryptocurrency’ with ‘Artificial Intelligence’?
“Could you revisit all the previous presentation slides and substitute the term ‘cryptocurrency’ with ‘Artificial Intelligence’?”
This heading, extracted from a comic strip by Benjamin Schwartz in The New Yorker, accurately represents Silicon Valley’s current trend of AI cleansing.
AI cleansing may seem like just another methodical process, yet it’s a sophisticated and multifaceted occurrence. It’s crucial for all individuals delving into this piece — including tech experts, advertisers, product developers, end-users, and IT practitioners from various backgrounds — to grasp the exaggerations, distorted emphases, and blatant fabrications that we come across not only in marketing and sales but also in the narratives we consume based on industry assertions.
Grasping AI cleansing
AI cleansing is a misleading marketing tactic that exaggerates the significance of artificial intelligence in the product or service being endorsed. The term draws its inspiration from “greenwashing,” coined by environmental advocate Jay Westerveld in 1986, where consumer goods are marketed as eco-friendly regardless of their environmental repercussions.
[embedded content]
Products utilizing traditional algorithms are branded as “AI-fueled,” capitalizing on the lack of a universally agreed definition for what constitutes AI and what does not. Startups develop applications that integrate with a publicly accessible generative AI API and market such apps as AI-driven. Expansive AI projects that are meant to demonstrate technology often rely on individuals working behind the scenes, as humans are the essential component to uphold the ambitious AI innovation.
Now, let’s delve deeper into the latter.
AI: A creation of human effort
The retail behemoth Amazon unveiled 44 state-of-the-art establishments named Amazon Go and Amazon Fresh, which (starting in 2016) leveraged the company’s suite of “Just Walk Out” technologies. (I initially reported on this initiative in 2017.)
Amazon’s vision was to establish stores where customers could walk in, select items from shelves, and calmly walk out without encountering a human attendant at a cash register. Sensors, including cameras, fed data to AI, enabling it to identify the buyers of particular items and bill them accordingly — all without a conventional checkout process. It gave the impression of pilfering, yet remained within legal boundaries.
The infrastructure was powered by cutting-edge computer vision camera systems, observing customers and their selections. Shelf sensors gauged the weight of items lifted, validating the products identified by the cameras in terms of type and quantity. RFID-tagged items contributed supplementary data. Advanced machine learning algorithms processed the data derived from cameras and sensors to recognize products and connect them to specific shoppers. Automated entry and exit gates monitored customer entrance and exit times.
The algorithms underwent training based on millions of AI-generated images and videos to identify products, human conduct, and human interactions.
For seven years, Amazon exhibited eagerness to elucidate the components of its Just Walk Out technologies. Nonetheless, the tech titan steered clear of elaborating on the 1,000 or so human workers recruited to facilitate the smooth operation — only acknowledging their role following media disclosures. Even then, Amazon obscured the specific functions performed by these employees, stating that they were not involved in video review.
Despite deploying 1,000 personnel overseeing and managing 44 stores (verifying three-quarters of transactions, per reports), the technology grappled with issues such as delayed receipts, mishandled orders, and elevated operational expenses.
This year, Amazon has been phasing out the Just Walk Out technology from its primary stores but continues to offer it as a service to external companies.
Another substantial instance of humans enabling AI in the background is encountered in the realm of autonomous vehicles.
Alphabet’s Waymo (formerly known as Google’s self-driving vehicle venture) operates a mission control center akin to NASA, where employees monitor vehicles via cameras and intervene remotely in case of incidents. (Here’s a time-lapsed video I recently captured during a ride in a Waymo car through San Francisco.)
General Motors’ Cruise subsidiary acknowledges the need for human intervention in their self-driving taxis every 4 to 5 miles, with each remote control session lasting approximately 3 seconds.
Several self-driving firms lean heavily on remote human operators. Notably, a German outfit named Vay explicitly employs human operators to steer the vehicles from a distance. The company recently rolled out a valet parking service in Las Vegas. The vehicle is driven to you remotely, and you subsequently operate it to your preferred location. Upon arrival, you disembark, and a remote operator undertakes the parking duty.
Amazon’s stores and self-driving automobiles exemplify a trend that is pervasive throughout the industry.
Reason behind AI cleansing
The top-tier, handsomely paid technocrats engaged in building AI systems place their faith in AI and its purported ability to tackle incredibly intricate challenges. In theory, it can do so. They relay this conviction to their superiors. Those leaders convey the same message to their boards. Corporate executive suites guarantee investors of this capability. And collectively, they assure the public of AI’s prowess.
However, there lies a minor hiccup: it’s an unattainable feat.
Most firms sense a degree of obligation towards their grandiose assertions, thus they obscure the extent to which their product or service hinges on human intervention behind the scenes to make crucial decisions, address challenges, and enable the supposed “magic” to materialize.
The more audacious entities remain unfazed by evidence refuting their AI’s actual capabilities as advertised or believed, opting to reiterate their claims time and time again. A prime example is Tesla CEO Elon Musk.
In October 2016, Musk proclaimed that Tesla would showcase a fully autonomous journey from Los Angeles to New York by the close of 2017.
By April 2017, he forecasted that within approximately two years, occupants could rest inside their vehicle while it autonomously maneuvered.
In 2018, Musk postponed his deadline for implementing fully self-driving Tesla vehicles to the culmination of 2019.
In February 2019, Musk pledged complete self-driving capabilities within the same year.
In 2020, Musk asserted that Tesla would unleash over 1 million self-driving robotaxis on the streets by year’s end.
Even in the current year, Musk hinted that fully self-driving Teslas might materialize “later this year.”
This anticipation will not materialize. Musk is deceiving both himself and his clientele. Musk epitomizes the epitome of AI cleansing.
The actual issue with AI cleansing
The collective impact of AI cleansing leads both the public and the tech sector astray. It fosters the fallacy that AI can accomplish feats that remain beyond its capabilities. It instills the belief that AI represents a universal panacea for all problems — or a potential descent into a dystopic scenario, contingent on individual perspectives.
AI cleansing incentivizes substandard solutions that prioritize “flair” over quality. Boasting that your pet-washing conduit is “AI-driven” doesn’t guarantee cleaner canines at the end. It merely implies you possess an excessively priced hose.
AI cleansing distorts funding endeavors. Presently, Silicon Valley investments predominantly gravitate towards authentic AI solutions and AI-washing alternatives. Even discerning investors might overlook the embellishments and falsehoods prevalent in AI-washing, secure in the knowledge that the AI narrative will resonate within the market owing to buyer gullibility.
Nonetheless, the most pressing dilemma doesn’t stem from the industry’s deluded sales tactics but rather from self-deception. Pioneers of AI solutions regard human intervention as a form of disgrace, whereas I believe the inclusion of human input would be received with appreciation.
Individuals indeed prefer human involvement in their shopping and driving interactions.
What the contemporary landscape necessitates is an infusion of more humanity and less automation. Presently, AI-generated inanity inundates the scene with cringe-worthy prose and falsehoods, coupled with bizarre and at times terrifying imagery. Google is overly zealous about replacing its search engine with an answer engine, potentially inundating pizzas with adhesive substances.
What the populace genuinely craves is a search engine that steers us towards user-generated content or at the very least, corresponds to a PageRank mechanism favoring human input and discerning AI-generated content.
The AI cleansing phenomenon is predicated on deception. It hinges upon the misconception that individuals want machines driving all facets of creation and control, a notion starkly contrary to reality. It prospers on the illusion that integrating AI into any domain automatically enhances its functionality, a fallacious premise at best. And it endures through the fallacious belief that human employment represents a retrogression in technological progress, a notion categorically incorrect.
Enough with the delusive AI cleansing narrative! Vendors ought to uphold transparency concerning AI applications. Simultaneously, consumers must demand tangible evidence that any AI incorporations in the products and services they invest in indeed furnish practical benefits.
On behalf of the entire tech community — comprising industry personnel, consumers, and tech correspondents — a resounding call is issued to Silicon Valley: Cease the gaslighting and present an authentic depiction of AI.
About Author
