Sunday, July 22

AI early adopters work with small budgets, but ramp up expected

LinkedIn Google+ Pinterest Tumblr +

Among early adopters, organizations have not truly deployed AI and budgets are being spent on smart services or developing the constituent technologies of AI, according to a survey by Constellation Research.

And AI budgets remain relatively modest, with over 90% of survey respondents reporting plans to spend less than $5 million in 2018.

These low budgets signal cautious adoption and deployment of foundational AI technologies for now, wrote researchers.

Still, that’s expected to increase as 60% of organizations expect to increase investment in AI by more than 50% in 2018.


“Because AI often delivers successes exponentially, Constellation expects AI budgets to continue to rise by more than 50 percent annually for the next four to five years as AI R&D yields bigger successes at an increasing pace,” researchers added.

The Constellation 2018 Artificial Intelligence Survey consisted of 50 C-level respondents from firms described as digitally fluent first movers, early adopters and fast followers across 12 sectors including finance, insurance, technology and telecommunications.

The survey also covered a wide range of size of firm by revenue, from less than $10 million (29%) to more than $1 billion (also 29%).

Contrary to trends in the financial services industry, it seems that back office is not a priority for surveyed organizations, which reported disinterest in developing AI in cost centres: 48% say they do not plan to invest in AI for finance, legal and administration.

The survey also scrutinized workforce talent: 80% of executives say their organizations need to hire additional human capital to implement AI solutions, and 72% obtain new talent for AI projects via recruiting.

“Taken together, these two trends have the potential to culminate in a talent war as more AI projects come online,” researchers said.


Keep track of companies trying to develop AI without the data scientist, like DigiFi and DataRobot


Other aspects of AI adoption the survey covered include the methods by which it is developed, maturity of AI components, the wake-up call of data privacy requirements, and the ways in which resistance to AI shows up organization-wide.

Top sources of resistance among those who report resistance to AI are: lines of business at 67%, IT at 32% and HR at 25%. Respondents cite trust (the inability of stakeholders to trust AI to perform tasks as specified), IT infrastructure and budget as the main reasons these departments resist AI.

For executives embarking on AI projects, Constellation has some advice about taking a step-by-step approach: start with a strong big data and predictive analytics foundation. From there, efforts will move to machine learning models and, eventually, neural networks.

“This effort could take more than two to three years to achieve. During this process,
organizations will train systems and codify contextual decisions for cognitive computing
with the goal of achieving a narrow AI. This effort will take more than a decade to
accomplish, but the basics must be in place for organizations to succeed,” according to the report.


About Author

Leave A Reply