AI ‘efficiencies’ not a euphemism for job cuts

AI ‘efficiencies’ not a euphemism for job cuts

AI ‘efficiencies’ not a euphemism for job cuts

by Ken Wieland

17 July 2019


LONDON — A recent
survey
from Gartner Research showed that many organisations,
unsurprisingly perhaps, use ‘efficiency’ targets to measure the success of AI
projects. But what does that mean in practice? Should employees feel nervous about
the axe falling on their jobs once bots get to work?  

Whit Andrews, vice president and distinguished analyst at
Gartner — and a specialist in use cases and business opportunities for AI and
cognitive computing — thinks efficiencies do not necessarily equate to job
cuts. Attrition of this sort, he told AI Business, was not high on the
corporate agenda of most organisations eyeing up AI implementations.

“In the current
economic climate in western Europe and US, my experience from talking to
clients is that they’re asking how I can achieve [with AI] the levels of
customer service I’m expected to achieve with the people I’ve got,” he said.
“Most organisations are going to look at AI and say that efficiency means I can
accomplish more things than I could ever do before.”

In the healthcare
sector, said Andrews, the drive to empower existing staff through AI and other
advanced technologies is particularly apparent. “I’ve never spoken to any
healthcare player which has characterised any decision they were going to make
as being promising because it would allow them to have fewer workers,” he said.
“I’ve only talked to them when they want to use technologies to make their
workers more successful.”

Andrews
acknowledges there will be some organisations not fully committed to the idea
that their people are major assets, but he rarely encounters them. “I never
talk to anybody who is saying I want to cut people,” he said. “I talk to people
who say I’m trying to cut training time; trying to leverage the people I have;
trying to find people faster; trying to reduce the necessary skills; and trying
to scale up.” 

There are some
nuances about what organisations mean by efficiencies, even if downsizing the
workforce is not on the cards. “Using efficiency targets as a way of showing
value is more prevalent in organisations who say they are conservative or
mainstream in their [technology] adoption profiles,” said Andrews. “Companies
who say they’re aggressive in adoption strategies were much more likely instead
to say they were seeking improvements in customer engagement.”


Related: Interpretable Automation Is The Future Of AI


The Gartner view is
that AI will result in net additions when it comes to jobs. While there will be
a ‘displacement effect’, removing or reducing the need for certain types of
jobs, there will also be a ‘productivity effect’ where there is increased
demand for labour to carry out non-automated tasks. Gartner calculates that the
productivity effect will more than compensate for the displacement effect in
terms of hard numbers, although the nature of future employment — if Gartner
is correct — will clearly change. Unskilled workers on this analysis look
especially vulnerable, unless they can be trained up for new roles.  

Mind the AI
skills gap

For all the AI
excitement about organisations doing more and scaling up, helped along by automated
processes, revamping the workplace to achieve these goals is far from
straightforward.  A worrying finding from
the Gartner survey is that more than half of respondents said that lack of
skills was one of the big challenges in adopting AI. Not really understanding
AI use cases was something that 42% of respondents fretted about. A
third of those canvassed reckoned the scope and quality of data was not up to
scratch.

“Finding the right staff
skills is a major concern whenever
advanced technologies are involved,” said Jim Hare, a research vice president
at Gartner. “Skill gaps can be addressed using service providers, partnering with
universities, and establishing training programmes for existing employees.
However, establishing a solid data management foundation is not something that
you can improvise. Reliable data quality is critical for delivering accurate insights, building trust and
reducing bias. Data readiness must be a top concern for all AI projects.”

Hare’s views were echoed by Rob Dalgety, industry
specialist at Peltarion, a Sweden-based AI software specialist. He emphasised
the need for ‘right partnerships’ that could remove some of the pain from AI
implementation.

“Organisations could deploy an operational AI platform that
takes away some of the core challenges in this area,” said Dalgety. “By giving
AI projects a graphical interface and abstracting above the underlying
complexity, using pre-built AI workflows and models with better integration
into IT infrastructure, organisations can reduce the cost, skills and
infrastructure required to run these projects, and move AI projects from
concept to production much faster.”