This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 3099067.
Ten Recommendations For Leveraging AI In Management
by Ciarán Daly
By James Smith
WASHINGTON, DC - If you work at the cutting edge of artificial intelligence (AI) innovation, you may believe there is no human cognitive ability which machines will not also possess - and that the rate at which this could happen may outpace our ability to prepare.
While general artificial intelligence by machines is interesting for humanity to consider, it is of limited practical concern for businesses today. The hype generated from daily news headlines of AI breakthroughs has been slow to translate into quantifiable business impact for all but a few corporate giants and tech companies.
Where are the reduced costs? Where are the increased efficiencies? And where are the disruptive opportunities? Substantive business opportunities to reduce costs, increase efficiencies, and create disruption do exist. But the best business strategy to capitalize on those opportunities is not an arms race to the biggest and most AI.
The best strategy is rooted in a re-evaluation and re-tooling of current business processes with an eye towards disruptive opportunities as technology evolves. Capturing value requires much more than AI technology or software, and success is, ironically, human centered.
Here we go beyond the hype and identify ten recommendations that boards and corporate management should consider as they prepare their companies to capture AI-driven business value.
1. Think of AI as a partner, not just as a tool.
Traditional computer technology, like the spreadsheet, is a simple tool; it accomplishes little more than its human operators and creates no new knowledge on its own.
AI differentiates itself by generating knowledge and insight, making it a partner (of sorts) that can collaborate to solve problems in innovative ways. Human-only solutions or machine-only solutions will generate linear results, but collaboration between the two can accomplish exponential results.
Some view the idea of human-machine partnerships as a frivolous personification of the machine; others see it as an important mindset that can help business move beyond traditional constraints to create something new (and valuable). Maintaining an open mind is an advantage.
2. Foster your company's data supply chain to fuel AI
Good data is required for AI to work - it is the fuel that powers AI. This makes data central to the modern company, and time spent on data-related activities will make up the majority of any AI efforts.
Companies will want to develop a formal data management plan or strategy that outlines how it will acquire, clean, share, integrate, curate, store, protect, and utilize company, customer, and public information.
The quality, variety, and volume of data will also be important in determining the type of AI that will help in particular problem sets; deep learning algorithms, for example, require a much greater volume of data than machine learning.
3. Become a learning organization open to continual improvement and change.
Companies that do not learn to adapt are often replaced by those that do; and in the current, disruptive technological environment, replacement is occurring at a much faster rate.
A corporate environment that maintains a constant and genuine dialog about change is a classic management capability, but it is particularly well-suited for the rapidly changing environment that technology is creating today.
Frameworks to do this are plentiful. Concepts like ‘design thinking’ work particularly well in allowing companies to put aside old methods and consider new ways of doing business. Fast failure is also important; investing years and large dollars into a bet-the-company transformation can backfire and destroy existing value.
An increase in smaller experiments with acceptable failures will help companies make the most of specific emerging technologies which can best help specific business processes.
4. Rely on domain experts to define the problem and interact with data.
The hype associated with cutting edge technology has naturally lead to significant investment, but it is too often spent on loosely defined solutions trying to solve a broad range of problems.
The largest companies have invested heavily in AI and are currently dominating the AI marketplace. But the significant research and development in AI over the past five years has not only helped those companies, it has propelled the entire industry forward. Companies can take advantage of that momentum.
Importantly, they do not need to reinvent the wheel. AI development today is significantly more democratized than prior technologies; technology giants are anxious to share—often without cost—AI innovation in order to expand markets. In addition, there are thousands of companies specializing in AI development.
Many of these companies fail each year, while others develop useful solutions that companies can acquire or hire for their own use. Separating the wheat from the chaff of AI service providers is a challenge, but less challenging than investment lost in development of numerous AI ideas that fail.
6. Corporate leadership buy-in is essential.
A commitment of leadership, even leadership that does not understand technical details, is required to make corporate AI succeed. Because utilizing AI is a whole of company process, not just an IT implementation, leaders with a vantage on the entire company are needed.
Those that can make difficult decisions for the company are needed when new AI solutions require disrupting an existing operating process or business plan.
Companies that are data intensive (all companies at this point) need sound data management practices and capable data scientists, but they also need to cultivate a data-science culture throughout the company.
The foundation of having a data-science culture is that everyone in the company can manipulate and understand data and should use basic scientific best practices: hypothesis, using appropriate methodology, clearly documenting results, expressing conclusions in terms of statistical significance or margin of error, and peer review.
Technicians are important but putting as much data as possible directly in the hands of the domain experts is also important to solving specific business problems.
8. Embrace complexity.
Traditional business practice and tools encourage the simplification (sometimes over simplification) of problems and advice. Data-driven AI and ever-increasing computing power enable problems to be considered in much more granularity.
Companies will benefit from reviewing their business processes to look for historical short-cuts. While many of these short-cuts may have once been necessary and practical, maintaining the status-quo given modern capabilities may represent lost opportunities.
9. Transparency and trust are very important.
An increasing reliance on machines in modern business processes implies an increasing trust in those machines. A high standard of transparency and ethics when it comes to AI is essential to build this trust and to unlock the full potential of the technologies.
Not all AI is created equal when it comes to transparency, and understanding how algorithms function can require a substantive understanding of mathematics and computer science. Output is often not proof enough and companies will want (or be required to provide) documentation to convince stakeholders to trust the new process.
Risk management concepts - a clear human function - continue to apply in an AI world. It remains human nature to trust fellow humans more than machines—so humans will remain in the loop and new rolls and activities explaining AI will emerge.
10. Human resources should consider workforce shifts.
There are many functions now performed by humans that are better suited for machines. That is a reality that companies should consider as they hire and balance their workforce.
Despite all the talk that machines will eliminate jobs, the evidence does not yet support it. What is much more likely is that new technology will change what humans do within the company, and the skills they require for top performance will also change.
Machine collaboration is not intended to replace humans, it is intended to enable humans to do more and use their uniquely human skills in more powerful ways. A focus on human aspects of technological change should not be counterintuitive to management - it is where substantive new value will be found.
James Smith is the Senior Managing Director at Ankura Consulting.