What AI and Intelligent Tool Projects are being Abandoned?

(and what does abandonment of these HR Tech projects actually mean?)


We surveyed 542 HR executives and subject matter experts to find out what AI and intelligent tools were passing or failing in real-world deployments.


Failure is part of the transition to any new tool. While that’s holding true for AI and intelligent tool projects we’re finding that more sophisticated deployments are likely to experience even higher failure rates.


The chart shown above is from our study of AI and intelligent tools in HR. It shows failure rates from our 90-day study of 542 respondents answering a 28-question instrument (study details are summarized at the end of this article). Make sure to click on the chart to see the data. The chart shows the number of projects that were ‘tried and quit’ versus the number of projects currently in operation. The most important thing about the chart is that it shows that failure, actual failure, is a part of learning to use intelligent tools. The more sophisticated and intelligent the tool gets, the higher the likelihood for failure in implementation.


While anecdotal evidence is that things are getting better, failure is part of the transition to any new tool and failure produces learning for both the vendor and the practitioner. As the vendor accumulates experience, it learns how to solve problems with more grace and effectiveness. The practitioner learns to profit from mistakes and gains insight into reducing programmatic risk.


The most important thing about the chart is that it shows that failure, actual failure, is a part of learning to use intelligent tools. The more sophisticated and intelligent the tool gets, the higher the likelihood for failure implementation.

Generally speaking, the failures seem to be caused by deep misunderstanding of what’s involved, how much it costs, and where responsibility lies for project success. Introducing any new level of automation changes the entire socio-cultural system of work in the customer organization. As the machine learns to cover the most basic or heavily administrative tasks, the work done by humans gets harder and more complex.


Because machine co-workers create the expectation that work will be done and answers will be provided at machine speed, this new and more complex work that the humans do has harsher deadlines and less of the relief that comes from doing a bit of simpler work.


Burnout rates increase and compensation creeps up to account for the new complexity. The worst of it is that training replacement workers becomes a new challenge. The traditional ‘on-the-job-training’ program boils down to giving a new person simple tasks to learn from and building their confidence and skill so they can slowly take on more complexity. Now, the machine does those simple tasks but doesn’t easily grow into the more complex.


What it actually means


2017-04-21 HRExaminer photo img sumser john bio pic IMG 3046 black and white full 200px.jpg

John Sumser is the Principal Analyst for HRExaminer.

We read the failure rate to be a measure of both new technology requiring sophisticated skills and a new mindset while nobody has had time to develop those skills because it is all new. The entire premise of current technical execution is that the way to learn is with customer participation, which means adoption is an important factor in effective use because that is the only way that people will learn how to effectively use these tools in practice. We expect to see the failure rates decline over time.


The survey points out several key things:

  • It’s very early in the adoption cycle. Almost all current customers qualify as early adopters.
  • Depending on the technology, between 30% and 70% of the potential market either doesn’t see the value in the tool or don’t even know what it is.
  • High failure rates in some areas suggest that its prudent to move slowly and ask a lot of questions.
  • There is growth already programmed into the market. That growth is dwarfed by the part of the market that doesn’t understand or can’t see the value. Massive education is required.
  • Intelligent tools require new and different kinds of management including data governance. Most HR Departments are unprepared.
  • The speed of market adoption and product refinement is astonishing. Most HR Tech markets grew at single digit rates in their first several years. It looks like intelligent tools are accelerating.



The Survey


The survey was designed to investigate three primary areas:

  • Which intelligent tools and technologies are actually being used and evaluated
  • How new technology is acquired and evaluated in HR Departments
  • The degree to which survey participants see new technology replacing existing workers


Over the course of a 90-day period, we collected survey responses from 542 individuals using a 28-question instrument. We used email lists to target HR executives. We did not purchase any responses. The breakdown of participants was as follows:


  • CHRO and Vice President of HR: 30%
  • Director Talent Acquisition: 12%
  • Recruiting: 17%
  • L&D: 9%
  • People Analytics 4%
  • Other 28%


Read previous post:
HRExaminer v11.00

We’ve got a whole week of amazing videos from John Sumser covering different impacts of the pandemic on the workforce.