- Where Insurance & Technology Meet

AI: Do It Easy or Do It Right

Artificial Intelligence is a prime mover for new opportunities in the insurance community.  From marketing and underwriting to risk and claims management, insurers are deploying AI functionality.  But casual implementations can eliminate value.  However, according to a recent McKinsey report, Canadian businesses are implementing with rigor and commitment.

The Research

The McKinsey & Company report – AI Looks North: Bridging Canada’s Corporate Artificial Intelligence Gap – was published in May 2018. McKinsey surveyed 120 Canadian executives and conducted in-depth interviews with 31 leaders. According to McKinsey, there was significant interest.

That said, the researchers found three  gaps in the results and, significantly, with the understanding of AI’s capabilities:

  • Strategy. While 89% of the business respondents said that “AI will create major, positive change within three to five years,” only one-third have altered their strategies to seize potential benefits;
  • Operations. 82 percent reported use / investment of AI apps, nearly all of the respondents were relying on traditional analytics;
  • Exploration.  Experiments to test transformation were underutilized.

The authors make the following recommendations:

  1. Become Fluent in AI.
  2. Reimagine the Business as a Tech Company.
  3. Build a ‘Digital and Analytics’ Backbone.
  4. Experiment and Scale AI Applications.
  5. Embed Analytics and Analytics-to-Business “Translator Talent.”
  6. Manage Human Changes During the AI Transformation.

Those damned humans again …

The first five elements are a bit challenging, but doable.  The last one, however,  is the bone crusher.  ‘Managing change’ is never easy, but for a somewhat traditional organization, the new world defined by Artificial Intelligence is more than a little scary.

McKinsey puts it well:

It should be obvious that the broader the adoption of AI across the enterprise, the greater the value a business will retrieve. There are, however, barriers that must be eliminated to achieve enterprise-wide adoption.

The authors note that barriers are as tricky as they get: “concerns about job loss, the fear of the ‘black box’ nature of AI (‘How does it work?’), and, most important, the fundamental human resistance to change.”

Take your time

McKinsey suggests that there must be “time in understanding how machine prediction will interface with human judgement.”

McKinsey notes thatAI has challenging limitations.  These include:

  • Data Labeling. Characterizing data properly.
  • Obtaining Massive Training Data Sets.  Machine Learning algorithms may require large aunties of data which may limit AI.
  • Explainability.  Difficulties in tracing and dissecting algorithm methods.
  • Generalizing.  Models trained for specific experiences may not apply to other situations.
  • Data and Algorithm Bias.  Misrepresentations of data will cause intentional or unintentional false results.

The McKinsey report concludes that AI has the capability of “unimagined value opportunities.”  However, “granular understanding of AI” is necessary to avoid small or large misinterpretations.

Leaving on a high note

McKinsey ends the report with a nod to the Great White North:

Canada has the will to succeed. There is no reason it shouldn’t be at the forefront of this new age of AI.

And, on a related note, Doug McElhaney, Associate Partner, McKinsey & Company will be presenting The Future Ain’t What It Used to Be at the 2018 Insurance-Canada Executive Forum on  28 August 2018.

What do you think?

AI is a critical component in the new insurance world.  I’d really appreciate your thoughts.