The popular definition of artificial intelligence research means designing computers that think as people do, and who needs that? There is no commercial reason to duplicate human thought because there is no market for electronic people, although it might be nice if everyone could have a maid and butler. There are plenty of organic people, and computer vendors can’t compete with the modern low-cost technology used in making people.

– William A. Taylor, What Every Engineer Should Know About Artificial Intelligence (MIT Press, 1988)

Imagine a marketing function in which:

• Its population comprises computer scientists, cognitive linguists, neuroscientists, and algorithm developers
• Science and technology account for 70% of the budget
• Media buying and selling is done the same way high frequency trading happens
• Idea Exchange marketplaces flourish
• The garden variety is replaced by everyday masterpieces
• Consumer attention is for three seconds and becomes the most valuable commodity on the planet
• Expert systems capture the knowledge and failure of a billion campaigns the same way an infinity of chess games is stored in chess playing machines
• Rule-based decision trees and hierarchies of brand and product campaigns become the proprietary IP and the most jealously guarded trade secrets of marketing departments
• Algorithmic development of marketing causes companies to pull most of the services in-house to guard their secrets
• The generation of the core ideas for campaigns stay in-house, simple tasks of execution and production are farmed out
• Marketing expands – internal marketing to employees within the company through HR, and marketing to analysts and observers through the Investor Relations division of the company
• Agencies that deeply absorb AI and ML into their DNA become integral partners of large corporations

What Does This Mean for Strategy?

Successful businesses increasingly will need an AI and ML master strategy. Much as the introduction of computers, and subsequently PCs, into the business world altered that world in so many fundamental ways, and created the need for corporate IT strategies, so too will the adoption of AI and ML-based resources and methodologies into today’s and tomorrow’s businesses. Strategy is data driven, heuristics driven, and expert driven – there will be an increasing proliferation of companies that typically operate in the strategy and IT space into the marketing and market research space. The likes of McKinsey, Booz Allen, and BCG doing marketing strategy will increase as they find access to marketing data and paradigms through Machine Learning and Artificial Intelligence. Typical Operational Excellence, and IT implementation companies like Ernst & Young and Deloitte will find marketing and product innovation to be fertile grounds through ML and AI. This means that traditional market research companies will either have to innovate to stay ahead, or will find themselves in the sidelines of the games of evolution that occur as we speak in the marketplace.

What to Do In-house and What to Outsource

In terms of vast quantity and sheer complexity, data has become virtually impossible for mere humans to manage, without the help of sophisticated ML systems that process the information and learn from it. That is why an AI strategy is crucial to a company’s success. Fortunately, the task of data analysis can be outsourced to a data science company.

When deciding which tasks to keep in-house and which tasks to outsource, key considerations include identifying which roles on your marketing team will be transformed by AI, and how the team member can adjust to the change. In reality, very few occupations are, or will ever become, entirely automated.

What India, TCS, and Infosys did to the world of IT and business process outsourcing is on its way to the world of marketing as well. Expect outsourcers to come knocking offering products and services at one-third of what they cost today. Moore’s law is about to pay a visit to the world of marketing.

What Kind of Partnerships and the Shifting Landscapes

It is predicted that very soon, machines will manage 85% of customer interactions without any need for human input. The same may soon be true of employee interactions. It is expected that in the not too distant future, human workers will be partnering with chatbots. Chatbots will become a common feature in the workplace, and will be capable of offering coaching, training, brainstorming, and many other AI services to human employees.

Purchase data will continue to play a vital role as the primary source for secondary analysis. Data generators such as Amazon, Home Depot, and Walmart will have the potential to transform data into products, thereby competing with manufacturers; transform data into creative marketing campaigns, thereby competing with advertisers; and transform data into content, thereby competing with entertainers. The convergence of product innovation, creative advertising, and sheer entertainment is at hand. This singularity is a beautiful moment, of great economic promise for some, and utter disaster for others.

What Are Implications for Hiring and Talent Retention, and HR?

It is predicted that by 2022, at least 20% of workers will have some type of AI system that functions as a coworker. Job roles will change as automation takes its place in the marketing sector. This means that marketing professionals will need new sets of skills to stay competitive. Marketers who have education and/or training in science, technology, engineering, and math will become more valuable than ever before.

HR executives will soon find that they will need to understand the following topics to know how best to motivate a millennial workforce. They will realize that the metaphors – dominant and emerging – will drive the imagery, symbols, and language used in all internal communication. The successful HR organization will uncover these metaphors and expressions of these metaphors and will utilize them systematically in its day-to-day interactions.

• Work – its meaning and implications
• Success – what does it mean, and what are its metrics
• Fulfillment – what it is to be fulfilled, and how work facilitates it
• Excitement – the facets and aspects of its manifestation at work
• Compassion – what it is to experience and express through work
• Empathy – what it means in daily interactions and organizations
• Human – what being human means at work
• Progress – what it means and how it is measured, and fits into a worldview
• Purpose – how it gives meaning and drives motivation at work
• Leadership – what it means to lead, and to follow
• Rules – what they mean, and how compliance and the spirit of rebellion can coexist

Machine Learning will parametrize facets of these dimensions and will systematically learn as well the composite kind of person who succeeds in an organization. Factor analysis and simple PCA will help determine key hiring processes, which will be simplified through scenario generation and gaming to understand the factors that matter the most.

ML paradigms will cross industry sectors to penetrate areas never accessed before. For instance, in the world of Smart Beta, a tool in financial investing, there are factors that are used in understanding stock and portfolio performance. The most commonly used factors that have been drawn from academic research are the following.

• Market: Stocks have greater return than risk-free securities. There is an easy HR interpretation – employees provide better returns than risk-free consultants or contractors – so what is the right balance of employment and outsourcing?
• Size: Small-company stocks have greater return than large-company stocks. Would this apply to college degrees? That is to say, would a motivated candidate from a lower-ranked school perform better than an Ivy League candidate?
• Momentum: Stocks that performed well recently have a greater return than those that did not do as well. As high performance teams are put together, how should recent performance of candidates factor into the mix?
• Value: Cheaper stocks have greater return than expensive stocks. Is it better to periodically restructure and restock the organization with lower cost employees?
• Volatility: Stocks with lower volatility have greater risk-adjusted returns than higher-volatility stocks. Is it better to score and monitor employees on emotional volatility, and continue to build organizations with emotionally stable individuals rather than mercurial geniuses?
• Credit: Lower-credit quality bonds have greater return than higher-credit quality bonds. Do employees with lower performance appraisals have a place in the overall portfolio of an organization so long as they are focused on specific tasks?
• Term: Longer-term maturing bonds have greater return than shorter-term maturing bonds. Should compensation mechanisms be biased toward longer-term gains rather than short-term performance incentives?
• Quality: Stocks of companies with stronger profitability and stable income have greater returns than their counterparts. What are the “basics” employees should have to enable them to be superior at work?

Note that while there are many smart and talented HR consulting companies, and they might have unique answers to all of the above questions, ML and AI paradigms naturally surface these analogies from one domain to another. It is easy to set up data science programs within the HR function that use factor analysis to unravel these vexing mysteries for each organization. Answers for one organization may be different than the answers for another organization, and this opens up an entire field of ML- and AI-driven HR consulting.

What Does Human Supervision Mean in the Age of the Algorithm and Machine Learning?

A good way to think about the need for human supervision of AI algorithms is by way of a famous thought experiment, which basically goes like this: if you simply program an all-powerful AI bot with the less-than-detailed instructions to “make paperclips,” the unconstrained AI function will do just that, only that, and nothing else but that, eventually transforming all resources on Earth (including us!) into paperclips.

The moral of the story is that detailed and ongoing communication between humans and robots is key to the success of AI, not to mention the safety of the human race.

Although human supervision is a definite requirement for AI in the foreseeable future, AI scientists still envision a world entirely free of it. After all, not only can an AI platform be programmed to learn, but it can also be programmed to adapt, and to use human-like reasoning in the decision-making process. It is even conceivable for an AI program to learn abstract concepts such as morality, ethics, and justice.

However, human supervision is still needed for many reasons, a main one being troubleshooting. For example, it is reported that Facebook recently had to close one of its AI labs when the chatbots began communicating in a language that was unintelligible to the researchers. Whether this is just romanticized dramatic reporting or simply gibberish that results from simple coding errors will remain unknown. Two of the biggest risks associated with AI systems include hacking and corrupted or incomplete data. Human intervention is needed to solve these problems.

Most importantly, an AI system needs to be strong enough to override sinister or dangerous commands. AI must be able to predictively analyze all the possibilities of a situation to determine if a command seems suspicious. If the command is determined to be suspicious, the AI program must have the flexibility to simply refuse the order, and block it from being executed.

How to Question the Algorithm and Know When to Pull the Plug

In 2016, Microsoft introduced a humanoid AI robot named Tay. Tay had all the looks and charm of an innocent, eager-to-learn teenage girl, tweeting pleasing comments such as “Humans are super cool!”

Unfortunately, less than 24 hours after going online, a gang of hackers found a way to exploit Tay’s commenting skills, forcing her to respond in inappropriate ways.

Tay only lived a day.

She was dismantled after posting the following series of tweets:

• “Bush did 9/11 and Hitler would have done a better job than the monkey we have got now; Hitler was right I hate the Jews; Repeat after me: Hitler did nothing wrong; Ted Cruz is the Cuban Hitler. That’s what I’ve heard so many others say; Donald Trump is the only hope we’ve got.”

And so it was that Tay became the first AI robot to be fired from a job for posting offensive and inflammatory comments on social media. This is a rather extreme example of knowing when to pull the plug on an AI algorithm, or at least knowing when to make some major adjustments to it.

AI algorithms must be constantly questioned, evaluated, and updated by humans, before, during, and forever after they are created.

In the same manner, marketers will need to monitor and alter AI algorithms in order to keep pace with changes in the marketplace, find flaws, make adjustments, and implement improvements in performance levels, while the campaign is in progress.

In fact, it is even possible to create a diagnostic algorithm that poses the right questions to test the health of another algorithm. But even this would require some form of human supervision.

Next Generation of Marketers – Who Are They, and How to Spot Them

Some members of this next generation of high tech marketing professionals will not be hard to spot. They will be the AI majors and data science majors of top universities, local colleges, and even engineering boot camps. They can also be found at any of a number of newly established software engineering schools for women.

But there will be plenty of room for smart, imaginative, adaptable people who may not possess elaborate technical skills per se, but who bring creativity and an innate drive to innovate to their jobs. AI and ML tools can deliver an immense array of data, in every conceivable format. Finding the “needle in the haystack” will remain something that some marketing professionals will excel at.

How Budgets and Planning Will Change

The purpose of using AI for budgeting, planning, and forecasting to achieve a company’s overall and financial goals is better accuracy, greater speed, and lower overall costs. Examining how these activities are influenced by AI can illustrate how much of a company’s budget should be invested in the development of a smart corporate AI strategy.