Telecommunications providers are more likely than companies in other verticals to use artificial intelligence and they aim to offer sophisticated services using AI, according to a recent survey by Infosys—but at the same time, they are more likely to be dissatisfied with their AI deployments than players in other sectors.
Telcos, according to the survey report, “still struggle with basic issues such as being able to clearly identify the problems for AI to solve. This means they are not applying AI to the right business problems—perhaps the reason for the low satisfaction.”
But, the report goes on, “there is a glimmer of hope for the industry. Telcos are better than average at data verification, ethics and bias management, and using deep learning … . But to truly lead, the industry needs to get better at identifying problems that AI can address, focus on simpler AI solutions and invest in AI infrastructure and compute resources.”
Infosys’ Data+AI Radar 2022 surveyed 2,500 AI practitioners from the U.S., U.K., Germany, France, Australia and New Zealand, across 12 major industries.
According to the survey, telecom players have been using AI longer than most; the majority of respondents said that they had deployed AI between 2-4 years ago (with 2020 being a peak year for AI deployment), a higher percentage than in other industries. Those capabilities are not particularly advanced at this point, however; 57% of respondents said that their AI is at the basic stages of development, which Infosys defined as “sensing” and “understanding”, meaning that the AI is capable of identifying patterns and detecting and making predictions; but that it still requires human involvement. More advanced development includes the ability to understand and respond autonomously to information, and to both respond and train itself to improve, a category that only 17% of telco respondents said their AI can currently achieve.
The survey further concluded that in the telecom industry, there is an “extremely high” rate of maturity on the use of AI and data in company business functions—but the rate of satisfaction with that AI and data use was low, especially compared to other sectors such as finance or retail and hospitality. In particular, the AI practitioners in the survey reported low satisfaction with AI tools used to suggest personalized recommendations or products.
The survey also compared the challenges that telecom respondents reported, to how long they had been working with AI deployments. The top concern among telecom AI newcomers, for example, was data verification, but the level of concern on that topic dropped the longer that the telecom respondent had been working with AI. However, telecom respondents who had been the longest-term AI deployments (more than five years), reported their top challenges as “clearly identifying the problems for AI to address” and AI infrastructure and related resources. Clear identification of problems for AI to solve was also an issue across other industries, with 22% of telecom respondents and 19% of respondents in other industries naming it as a top challenge. But the concern in telecom about infrastructure and resources, five years into AI deployments, was an outlier—in other industries, that concern was mitigated over time, according to Infosys’ report.
The telecom industry “could benefit by perfecting less complex deployments before opting for complex deployments,” the report concluded. “This way, it could invest in the right AI infrastructure and resources. Collectively, this could improve its satisfaction levels related to deployments.”
For more detail, you can access the report here (pdf).
