AUG 02, 2022
What obstacles does government AI still have to overcome?
OCT 10, 2021
Artificial intelligence made few gains during the pandemic, Gartner finds, even as more agencies turn to chatbots. Confusion about the technology and anxiety among government workers are among the main hurdles.
While public agencies continue to deploy chatbots and other artificial intelligence tools, confusion about the technology abounds, according to new survey findings from Gartner, and the pandemic has provided little fuel for its growth.
The research agency found that 36 percent of survey respondents plan to increase AI and machine learning (ML) investments this year.
Even so, proponents of AI and ML have significant work to do, said Dean Lacheca, Gartner’s public-sector research director, in an email interview with Government Technology.
“What was striking is the fact that leveraging of AI within government had not really advanced during the period of the pandemic. It hasn’t really slowed but it has not accelerated either, especially if we exclude chatbots from consideration,” he said. “There is still a lot of talking about the impact it will have and some experimenting and some great point solutions, but adoption is still not widespread. The research points to the use of ML ramping up in the coming couple of years.”
That’s not all that faces backers of the technology.
Gartner found that while 53 percent of government workers who have used AI tools say the tech “provides insights to do their job better,” only 34 percent of workers unfamiliar with AI said the same.
“I think there is still more work to be done to demystify the technology. There remains a lot of misunderstanding and generalization of the technology,” said Lacheca.
Part of the problem stems from a lack of specificity, he said.
“The more that government technology leaders start to identify specific and narrow use cases and then link them with the specific, readily adoptable technologies like ML, computer vision and natural language processing, rather than the generic AI terminology, the more likely the government leadership will be to understand the potential of the technology,” Lacheca said.
The Gartner findings stem in part from a global survey that attracted 166 responses from all levels of government, with 27 percent coming from state and provincial governments, and another 27 percent from local governments, as well as some respondents from counties. The findings also come from a separate Gartner survey of 258 government employees working for public agencies around the world.
Gartner found that chatbots stand as perhaps the most popular or familiar tool in this area of gov tech, with 26 percent of respondents saying they have deployed them, and 59 percent expected to within the next three years.
Even though AI and related tools can still seem futuristic even now — and still have a way to go before they are mainstream in government — some of the technology is accessible to public agencies of various sizes, and not just relatively well-funded federal government units, according to Lacheca.
“The reasons why chatbots have taken off in government is that they are pretty much package solutions, with only minimal concerns about data privacy, which can offer a perceived speed to value for government,” he said. “So anywhere that an off-the-shelf or easily configurable/trainable AI solution can be stood up rapidly with little data privacy concerns are the likely areas of rapid investment by government.”
It’s not only chatbots that promise to find more use among public agencies. Data mining enabled by ML has been deployed by 16 percent of survey respondents, with 69 percent planning to do so within the next three years, Gartner found.
What Gartner called more “specialized” AI tools also could soon find more uses within government — including local agencies.
Such tools include geospatial AI, which, according to Gartner, uses “(AI) methods to produce knowledge through the analysis of spatial data and imagery.” Those tools will by their nature have lower rates of uptake than chatbots, but are likely to be deployed in such areas as defense, intelligence, transportation and local government.
Along with the spread of AI will come anxiety among public-sector workforces, according to Gartner.
The research firm found that while 42 percent of government employees who have yet to work with AI believe the tech helps get work done, only 27 percent of those respondents “believe AI has the potential to replace many tasks.”
Meanwhile, 31 percent of employees who have used AI view the technology as a job threat.
Even so, 44 percent of respondents who have used artificial intelligence think the technology improves decision-making, with 31 percent saying AI reduces the risks of mistakes.
That said, 11 percent of respondents think AI makes more errors than do people.
“Senior executives in the public sector must address the early apprehension among the government workforce by showing how the technology helps them to getting their work done, then continue to build confidence in the technology through exposure, use cases and case studies,” Lacheca said.
Natural Language Processing