What AI Cannot Do (and what humans can)
Much has been written and said about AI, all of the tasks it can achieve. Frequently, almost in awe, the narrative is focused on all AI can do and the changes it will wrought. The negative mostly confined to the inevitable job loses, and resulting social upheaval. Then there is some talk about the future of AI, where will it go? We are very much in the infancy of AI introduction. (Stanley Kubrick’s 1968: 2001 A Space Odyssey and HAL, comes to mind.)
Some people get carried away, thinking AI will “overtake” us, determining we are not only no longer needed, but pose a potential existential threat, so they “wipe us out”.
Woah, slow down Sherlock. They are just machines, not a life force. Understanding what AI can, and cannot do, highlights where and why people are needed. I am old enough to remember when computers first became common in the workplace. In 1980, when I started working in market research, there were no “word processors”. Instead we had a typing pool to type our reports. Quantitative research analysis was done off-premises, by a mainframe computer. The questionnaires were hand coded and punched cards used to program the mainframe. At the time there was a large level of concern about job losses computers in the workplace would cause. AI is the same phenomena as this, on steroids.
The result was not thousands of suddenly unemployed people on the street begging. Society adapted and computers were absorbed into the workplace, with the only major effect being a huge increase in productivity.
So what are the 2 major areas where AI cannot replace people?
1 - MATHEMATICS
The digital world has give us tons of data, and with-it a new profession – data scientists, working with numerous computer data analysis and statistical programs (SPSS, R, MATLAB, SAS etc) to analyse it. Prior to computers, we used slide rules and much work done by hand. Anyone who is a pure mathematician, from which statistical formulas are derived, will tell you these programs make a ton of assumptions about the data fed into them. Having studied pure and statistical maths, we always derived the PDFs, from scratch, rather than use “off the shelf” statistical formula.
AI has added to the volume (the number and type) and the speed, which statistical analysis can be conducted – SPSS was the dominant program back then.
I asked ChatGPT to list all of the statical functions and analysis AI can undertake. At first glance, very impressive. (Or boring as batshit, depending upon you bent for maths).
Data Collection and Preprocessing
Exploratory Data Analysis (EDA):
Statistical Analysis
Machine Learning Models:
Implications of AI-Driven Analysis:
AI’s ability to automate and enhance statistical analysis enables businesses to derive actionable insights from consumer data, driving strategic decision-making and improving customer experiences.
Yet for all AI can do, 2 words stand out “actionable insights”. Or more bluntly, all of the above information that AI provides “ain’t worth a cracker”, unless people running the business or organisation, know what to do with it. Computers on steroids.
When asked about Pure Maths and AI’s ability:
“As of now, humans are generally better than AI at solving difficult pure math problems, especially those that require creativity, intuition, and novel insights.
Statistics and Physics/Applied Maths are the applications of pure maths. Pure mathematicians work on “problems of nature and space”. Often their findings have no current application. Most would have at least heard of Calculus – attributed to Sir Issac Newton, though the German mathematician Gottleib Leibniz, developed it around the same time. Newton first published his work on calculus in 1687, yet it was Alessandro Votto who first discovered a steady flow of electrical charge in 1800. (Benjamin Franklin’s kite experiment was in the 1700s but had no idea how the flow of electrical charge occurred). It was calculus that helped to explain how electrical charge flowed, nearly 150 years later. (Continuous distributions),
There is no need for high-level pure maths in advertising. BUT AI does not interpret and determine strategy, be it financial, marketing, advertising, or financial engineering. There are similarities to the introduction of computers to business and the introduction of AI. More and more “grunt work”, some of it very sophisticated grunt work, will be done by computers/AI, but the Thinking and the Imagining can only be done by humans.
2 - CREATIVITY
AI and its use for creativity in advertising is where many in the industry “shit themselves”, thinking they will be made redundant. I find it amusing when so many today write about how advertising has changed, some say radically, from pre-to post-digital. I try to read as many relevant articles (and watch occasional videos) in an endeavour to stay up to date with what is happening in the industry. Recently I came across a presentation I gave in 1997 to Toyota Dealer Principals. Though much in the media word has changed, the all-important basics have not: What are your objective/s, who are you talking to and what do you want to say?
AI is not just about computers. If it is “independent” thinking, biology comes into play. To independently think versus follow sophisticated programming, then it must be a “life force” of some type. AI doesn’t formulate opinions – its conclusions are not based on thinking as we know it, rather the data it is fed. There are no amino acids, RNA, DNA etc.
The recent passing of Jack Vaughn gave me pause to think about creativity and AI. Many readers would be too young to remember the Campaign Palace. The Campaign Palace was named Campaign Brief Agency of the Year in 1987, 1991, 1992, and 1998 “And would have done so almost every year from1972 to 1986 if it had been around” former Palace CD and fellow creative legend, Ronnie Mather said. The agency won every major award worth winning at every major award show, was respected around the world and put Australian advertising on the map. I was fortunate to work at Saatchi & Saatchi in the late 80s (before the M&C split), which was a revolving door to the Palace. (Ronnie Mather was CD at both the Place and Saatchi.) Bob Isherwood took over the Saatchi CD role after Ron left (he was Ron’s art director). These were the days when a creative department was made up of teams of writers and art directors. (Bob ended up Saatchi’s worldwide creative director.) The Monkeys have been praised as the best creative agency in the country, by a long shot, for nearly a decade. But comparing them to the Palace is not possible. Their era and that of the Monkeys are 2 completely different worlds. Advertising has always been very liberal, but Woke it was not.
Planning came into the Palace, but its objective was to help enlighten, rather than strangle the life out of creativity. There were no “content creators”, influencers or the myriad of all the other tasks and titles which now make up the advertising/communications industry. Creative was its product and prided itself on being “The Best of the Best”.
As long as I can remember (over 40 years), advertising has been described as:
“Where science meets art”
AI adds a ton of science to the game, maybe to the point where it dominates. We are almost drowning in data. Bernie Taupin, the song writing partner of Elton John summed up AI well:
“Lots of people ask me how I write songs. I don’t know. AI cannot write songs as well as the human heart can because it has no heart. I loathe and detest the whole idea of AI and creativity.”
I personally experienced the change and uptake of data analysis and statistics in marketing and advertising. Way back in 1978, I had a 3-hour statistical maths (not statistics, but the mathematics of statistics) lecture from 6pm to 9pm on Friday nights for a semester. (The day lecture, earlier in the week, clashed with a biology lecture). In 1978 we did not have the luxury of being given typed lecture notes. The lecturer talked, using a white or chalk board at random, and students wrote like crazy. Any questions. That was for tutorials – asked of the tutor. That Friday, I had tickets to a Bob Dylan concert, an incredibly rare event from a genuine legend. I asked the lecturer, Professor David Chant, could he please pass on notes to the tutor, as I would miss the lecture. Professor Chant was aghast, and I can still hear his response. “You and music Strohfeldt. You will end up working in a, in a (he was struggling to think of a job far removed from mathematics), in an advertising agency!”
How times change.
AI can pump out “quite acceptable” creative, consistently at a massive rate, with slight to even quire major changes to appeal to a wider variety of targets. But AI cannot think, it cannot imagine, it cannot “create”. That will, for the foreseeable future, be a human task. The day AI starts to create, keep an eye out for HAL.
