Despite concerns over data privacy and immature technology, AI assistants and chatbots are gaining a foothold among businesses, according to a report from IT professional network Spiceworks.
The report, which surveyed 529 IT professionals from North America and Europe, showed that 29% of respondents have implemented one or more chatbots or intelligent assistants for work-related tasks, or plan to do so this year.
Adoption was higher for larger organizations, which have more resources to evaluate emerging technologies, with 24% using AI assistants or chatbots currently and 16% set to deploy the technology this year, for a total of 40%.
“A lot of businesses are starting to see how these technologies are useful, and because they are more widely available, we are seeing adoption increase,” said Peter Tsai, Spiceworks senior technology analyst and the author of the report. He added, however, that the technology is still “in its infancy.”
How are AI assistants being used?
AI assistants and chatbots are used in a variety of ways, the most common of which is voice-to-text dictation (cited by 46% of survey respondents), while 26% have used them to support team collaboration tasks and 24% deploy them for employee calendar management. Other use cases include email management (14%), customer service (14%), IT help desk management (13%) and data analysis (10%).
The report also indicates where AI assistants and chatbots are being deployed within organizations. IT departments account for the majority of use cases (53%) — perhaps unsurprising, since IT pros tend to be early adopters of new technologies and may be testing the tools ahead of wider business deployment.
The IT department is followed by administrative and business management (23%) and customer service and support (20%). Marketing and sales — both 16% — are also using the technologies, while there are low levels of adoption in accounting and finance (9%), research and development (7%) and human resources (7%).
Cortana leads in the workplace
While Microsoft’s Cortana is lagging in the consumer market, it was found to be the most widely used AI assistant in the workplace, thanks to its integration into Windows 10. Of the respondents that have already implemented intelligent assistants or chatbots, 49% are using Cortana, while Apple’s Siri — which is built into iOS and macOS — is used by 47%. Google Assistant is used by 23%.
Amazon Alexa has made less of a mark, despite being the most popular consumer AI assistant. Only 13% of respondents are currently using Alexa, though a further 15% of organizations plan to use it in the next 12 months. As well as announcing a partnership with Microsoft that will see Alexa integrated into new Windows 10 laptops, Amazon has launched its Alexa for Business service aimed at boosting enterprise adoption.
Some businesses may have been put off by Alexa’s initial positioning as a consumer-focused assistant built into smart speakers for the home, Tsai said. “There was a conversation about a year ago where people were concerned about privacy, especially in a business setting, with an app that originates from Amazon. People are concerned about the always-listening features with some of these intelligent assistants,” he said. “That might also be a concern among IT professionals.”
Chatbots are also making their mark in the workplace. Fourteen percent of organizations are using chatbots within collaboration platforms such as Slack and Microsoft Teams, and adoption is likely to increase significantly over the coming year, with a further 16% expected to adopt the tools. However, businesses appear to be less keen on creating their own custom chatbots, with only 2% having done so; 10% plan to do so in the next year.
Barriers to adoption — privacy concerns and immature tech
While there is interest in business applications for AI assistants and chatbots, there are various drawbacks to the technologies that are holding back wider use, the report said.
The primary reason for not implementing these AI systems is their limited use cases, cited by 50% of respondents who said they do not plan to adopt the technology.
This is likely because the technology is “still evolving,” said Tsai. For example, the top complaint among chatbot and intelligent assistant users is that the software “misunderstands what you are trying to tell it, or it doesn't understand nuances in human dialogue, slang or jokes, or colloquial speech.”
This makes businesses wary of investing in the technology. “If you have to put more effort into the technology to get it to do what you want, you are not really getting a good return on your investment,” he said. “It seems that many people will adopt intelligent assistants and chatbots in the future, but maybe right now the technology is not quite there yet.”
Twenty-nine percent of respondents said concerns around security and privacy are holding them back from implementing AI systems. This is particularly the case in regulated industries such as healthcare, where organizations are dealing with sensitive information, said Tsai. “There are a lot of regulatory concerns there. You don't want to divulge details and have them saved by a third party on someone else's servers, which would obviously violate privacy regulations.”
Such concerns have been highlighted by enterprise IT leaders. Speaking to Computerworld in January, Tom Cullen, CIO at Driscoll’s, the $3.5 billion supplier of fresh berries, said, “The main drawback [to virtual assistants] is security. The reason I don’t have one in my house is that it is always listening, and you don’t really know where all of the data is going and what is being done with it.”
Joel Jacobs, vice president, CIO and CSO at MITRE Corporation in Bedford, Mass., also questioned where data created in virtual assistant interactions would reside. “For example, if the voice interpretation is ‘in the cloud,’ does that mean that the voice track and transcription is being stored by the service provider? If so, how can they be used?”
Other reasons organizations have chosen not to implement AI assistants and chatbots, according to the report, include cost (cited by 25% of respondents), a fear of employee distraction and lost productivity (19%), accuracy concerns (15%), a lack of support from management (14%) and training requirements (13%).