BETA
This is a BETA experience. You may opt-out by clicking here
Edit Story

When It Comes To Artificial Intelligence, Throw Away That Tired Old Tech Rulebook

Following
This article is more than 4 years old.

If you’re in a relatively technologically advanced organization, you already understand how new technologies demand —or facilitate — new approaches to problems and opportunities. But when it comes to artificial intelligence, you ain’t seen nothing yet. AI opens up issues — such as risk and explainability — that most business and technology leaders have just begun to grasp. In addition, AI will be the determining factor in shaping the winners and losers in the market ahead.

Many of these points were brought out in an in-depth study of 30 companies leading the way with AI, published by KPMG. For starters, the number of employees directly in involved with AI work will expand over the next three years, from an average of 375 to about 600, the KPMG authors predict. They expect to double their investments in AI talent and technology during this time as well.

The profound impact of AI on enterprises was also explored in a recent survey of 2,280 business leaders from MIT Sloan Management Review and SAS, which finds that while we’re still the early stages of AI, it’s not too soon to start considering the organizational and risk implications.

The authors of these two studies point out the ways AI is poised to turn enterprises upside-down:

Business leaders will need to get directly involved in AI initiatives. In the KPMG survey, 40% of respondents indicate line-of-business executives will be taking leadership roles as well in AI initatives.“AI demands more collaboration among people skilled in data management, analytics, IT infrastructure, and systems development, as well as business and operational experts,” the report states. “This means that organizational leaders need to ensure that traditional silos don’t hinder AI efforts, and they must support the training required to build skills across their workforces.”

The CIO and CTO will have to get really involved. CIOs and CTOs will be taking leading roles in AI initiatives, both the KPMG and MIT surveys show. These executives “will need to prioritize developing foundational technology capabilities, from infrastructure and cybersecurity to data management and development processes — areas in which those with more advanced AI implementations are already taking the lead.” AI will change the roles of CIOs, who “will also need to manage the significant changes to software development and deployment processes that most respondents expect from AI. Many CIOs will also be charged with overseeing or supporting formal data governance efforts.”

AI will require an increased focus on risk management and ethics. Advanced enterprises are only just starting efforts to solidify governance, control and accountability to manage the risks with AI, the studies find. Only 25% to 30% of companies in the KPMG study are investing heavily in developing control frameworks and methods to drive greater trust and transparency in AI.

The MIT survey also finds “few practitioners have taken action to create policies and processes to manage risks, including ethical, legal, reputational, and financial risks,” the report states. “Those with more advanced AI practices are establishing processes and policies for technology governance and risk management, including providing ways to explain how their algorithms deliver results.” The survey finds about half are acting to create organizational structures to manage AI risk: 26% have a group that sets policies and manages AI risk, and 24% plan to create one.

AI risks identified in the MIT survey include the following:

  • Deliver inadequate return on investment
  • Produce bad information
  • Be used unethically
  • Support biased, potentially illegal decisions
  • Produce results that humans cannot explain
  • Be too unpredictable to manage adequately

Explainability will become critical. There already is some action on explainability, which the MIT survey report defines as “the ability to identify the key factors that a model used in producing its results, which may be recommendations in a decision-support system or actions in an automated process.” A majority of survey respondents with the broadest AI implementations, 55%, make explanations available to internal stakeholders, and 42% make explanations available to external parties such as customers.

Follow me on Twitter