Employees want more training on AI tools before using them for work
Many have tried out ChatGPT’s conversational AI for business but feel they need more information before making it part of their jobs.
Artificial intelligence is held up as a way to automate time-consuming business processes, but 49% of employees say they’ll still need more training to be able to use ChatGPT and other similar AI tools effectively, according to a survey conducted on March 7 by TalentLMS. The company, which offers an employee training software platform, surveyed 1,000 employees in the U.S. on the impact of the natural language AI solution, ChatGPT, and natural language AI as a whole in the workplace.
SEE: Artificial Intelligence Ethics Policy (TechRepublic Premium)
Workers find AI has a learning curve
Of the employees surveyed, 49% said they need training on how to use AI tools before they can use them effectively. Of the remainder, 23% felt they did not need training because AI tools were easy to use, and 14% didn’t plan to use AI tools at all.
What about using AI to train people on how to use AI? TalentLMS found that almost half of employees (46%) would trust learning content produced by AI. Another 27% of respondents say they wouldn’t, while a further 27% were neutral. Another 14% said they have already received training on how to use AI platforms at work.
“People may feel threatened by a technology that shows creativity,” commented Thanos Papangelis, TalentLMS co-founder and CEO of e-learning company, Epignosis. (TalentLMS runs on Epignosis software.)
“Conversely, productivity will get a huge boost, and our economies that seem to flatten will witness a boost similar to the internet or mobile phones. Jobs may be destroyed, but, based on history, much more jobs will be created.”
How do workers use ChatGPT today?
OpenAI’s ChatGPT has already been integrated into workplace software by some organizations, such as Salesforce. Of the survey respondents who have already used ChatGPT for work purposes, more than half of them (51%) said they still need training on AI tools. Another 18% have already received ChatGPT training. (Note that 70% of the total sample was in the group that had already used ChatGPT for work.)
Most of the surveyed employees who use ChatGPT have used it for writing content (36%), analyzing data (33%) or providing customer support (30%). Some of the lesser-used but potentially more impactful responses included “brainstorming and developing new ideas” (27%) and “navigating tough conversations” (22%).
One in five employers disapprove of AI for work
Most employees surveyed (70%) have at least tried out ChatGPT for work purposes. The remaining 10% have used it for personal reasons outside of work and 20% have not used it at all.
Employers have their own perspective on this: One in five (19%) employees who have used ChatGPT for work say their employer doesn’t approve. Employers are concerned about removing the human touch from certain tasks (48%), providing misinformation or factually incorrect information (44%), plagiarism (29%) or grammatical errors (27%). Employers may already know that ChatGPT’s search function is capable of returning coherent but factually incorrect responses
On the other hand, some may see it as an efficient way to save time on writing emails, meeting plans or other content. Employees who have used ChatGPT say it has helped them improve their time management (61%), boost their productivity (57%), access information and make decisions faster (53%), and increase confidence in their day-to-day role (37%).
What business leaders should keep in mind when adopting AI
Business leaders who are considering deploying AI should think about information security; for instance, if employees paste confidential company information into a ChatGPT chat window, that data could mix into the tool’s training dataset and appear in front of users at a different organization. Some companies, including JPMorgan and Verizon, have blocked access to ChatGPT because of this possibility.
Read next: As a cybersecurity blade, ChatGPT can cut both ways (TechRepublic)