There’s an emerging industry-wide and cultural consensus: intelligent computing has moved well beyond the sci-fi realm of The Matrix or The Terminator. Since 2014, Bill Gates, Stephen Hawking, and Elon Musk have mainstreamed the prediction that humans can achieve human-equivalent artificial intelligence – and probably within our lifetimes.
Under the assumption human-IQ-level artificial intelligence is inevitable, it would require perhaps decades of research and development and billions in investment. Who’s going to cover the R&D pricetag?
From all indications, advertising dollars – made possible by massive user bases and increasingly sophisticated analytics – could foot the bill. Consider Google’s $400 million acquisition last year of DeepMind, a London-based AI company, or Twitter’s purchase last summer of a Cambridge, Mass.-based AI startup. With combined user bases in the billions, ad-revenue driven social media and search engine companies have among the strongest business incentives and best available resources to pave the way for AI. How better to study the human mind than to study the behavior of billions? And how better to monetise that behavior than to replicate it?
How a supercomputer could sell products: Artificial Intelligence as a service
Data is the new oil, as the advertising industry has said for years. Social media posts and behavioral advertising could pave the road to artificial intelligence by providing what are perhaps the best insights into the factors behind real-world behavior. It’s no surprise to industry observers that Twitter, Facebook and Google are in an artificial intelligence arms race.
Today, AI in advertising is evolving rapidly, even for creative agencies. A recent example is a summer campaign with ads that adapt to user reactions, launched by M&C Saatchi, Clear Channel and Posterscope and dubbed the “first ever artificially intelligent poster campaign.” Using an intelligent algorithmic approach, the poster selects ads based on efficacy and continues to improve over time.
As one advertising executive wrote in 2013, “algorithms are already the audience”: the vast majority of stock trades are executed by computers, and from social media to search engine marketing, campaigns are already catering to computer programs, which in turn cater to real people. Predictive analytics seeks to anticipate human behavior – providing a significant business incentive to re-create it first. Facebook’s recent debut of M, an AI-driven executive assistant, is designed to ultimately make purchases on behalf of users – essentially promising to replicate and influence the behavioral factors behind their purchase decisions.
Certainly, artificial intelligence is of interest to virtually any company, regardless of industry. The most well-known super computer, IBM’s Jeopardy-winning Watson, is driving initiatives from a machine-learning focused cancer research programme to a computing initiative to help veterans transition to civilian life. McKinsey & Company is actively considering – and consulting – around how machine learning and cognitive computing can replace managerial-level employees.
McKinsey previously cited “automation of knowledge” work as the No. 2 largest factor unlocking $33 trillion in technology-driven economic value by 2025. On the consumer front, Mattel joined the fray in September with the introduction of “Hello Barbie,” an early-stage AI toy that aims to have fluent, Siri-like conversations with children.
Nevertheless, advertising technology – inherent in search engines and social media – is perhaps best positioned to provide the insights that will drive the progression of artificial intelligence. Not only can digital networks provide us with massive data sets for how to understand human behavior, but advertising currently provides perhaps the best funding source to reach for the Holy Grail of AI. This is further evidenced by companies like Google investing tremendously in scientists, research and companies.
What’s at stake: the promise of computing “superintelligence”
There’s today’s cognitive computing and tomorrow’s human-equivalent artificial intelligence, perhaps defined by the infamous Turing Test. Then there’s the concept of superintelligence. A not-insignificant number of scientists and academics forecast that the IQ of a single supercomputer could one day exceed the combined thinking power of every human on earth. This scenario is optimistic when we think about computers having the potential to one day cure cancer or problem-solve world hunger using unimaginable processing power. On the flip side, Elon Musk and 1,000 notable experts and public figures argue that artificial intelligence could be more dangerous to humanity than nuclear weapons.
How could the road to AI – and further, superintelligence, be possible? The 1970s-era computing term, Moore’s Law, forecasted that processing power doubles every two years, a prediction that appears sustainable long-term, thanks to advancements in semiconductors by companies like IBM.
On a purely mathematical level, Moore’s constantly doubling computer power prediction is a big deal. Think of the parable of the man who, after winning a chess game, asked a king to place one grain of rice on a chessboard and double it for each of the board’s 64 squares. The king gave up: had he completed the task, the 64 chessboard squares would result in a Mt. Everest-sized pile of rice, or enough to put all of India knee-deep in rice.
Apply this maths equation to advancements in computing power, and Musk, Hawking and Gates’ conclusions on our path to AI become more convincing. Think of it this way: some of the best minds in humanity are working on making computers smarter. If computers became smarter than humans, how fast could they advance their own intelligence?
Superintelligence’s capabilities are, by definition, unfathomable. But today’s advancements in cognitive computing already present a staggering trend line: if the experts are right, these exponential computing increases could mean a computer the size of a USB drive might be able to nearly seamlessly replicate a human’s thinking.
At what point could advertising campaigns be based on simulations using artificial versions of our mind – or the AI technologies we allow to buy for us? It’s no surprise that marketers are starting to encourage ad agencies to invest in cognitive computing technologies.
Today, a predictive analytics algorithm might guess the advertising spend necessary for someone in your general demographic to make a purchase decision. Tomorrow, machine-learning might allow a campaign to individually predict with unprecedented accuracy how you, among millions of others, will make your next move. If advertising-driven companies foot the research bill, your online search engine and social media activity could be fueling the rise of the machines.
By Joe Hyland, CMO, ON24
comments powered by Disqus