“It would be a mistake not to harness the power of this technology. There are many opportunities in terms of productivity and functionality,” says founder of an AI management software company.
Talent search platform Plum sees ChatGPT making waves within and beyond the tech industry, looks to the source to explain how staff can and can't use generative artificial intelligence chatbots I decided to.
ChatGPT, which can turn simple text instructions into poems, essays, emails, and more, produced a number of draft documents last summer, and the Kitchener, Ont.-based company is about 70 per cent of the way to final policy. did.
“There was nothing wrong with it, nothing out of the ordinary,” said Plum CEO Caitlin McGregor. “But there was an opportunity to be a little more specific or a little more customized to our business.”
Plum's final policy, a four-page document based on a draft of ChatGPT and compiled with advice from other startups last summer, tells staff to keep customer and proprietary information out of AI systems and to ensure that the technology He advises checking everything you spit out for accuracy and attributing the content your technology generates. .
This makes Plum one of several organizations in Canada to codify its stance on AI as people increasingly rely on technology to be more productive at work.
Last fall, the federal government released a set of AI guidelines for the public sector, many of which spurred policy development. Many startups and large organizations are now reworking these or developing their own versions to suit their own needs.
These companies said their goal is not to curb the use of generative AI, but to ensure that employees feel fully empowered to use AI responsibly. I am.
“It would be a mistake not to harness the power of this technology,” said Niraj Bhargava, founder of Ottawa-based AI management software company Nuenergy.ai. There is an opportunity.”
“But on the other hand, using it without guardrails poses many risks. Some are existential risks for the planet, but there are also real risks such as bias, equity, and privacy issues.”
Balancing both is important, but Bhargava said there is no “one-size-fits-all” policy that applies to all organizations.
If you're a hospital, you may have very different answers about what's acceptable than a private technology company, he says.
However, there are some principles that frequently appear in guidelines.
One is that we don't incorporate client or proprietary data into our AI tools. Companies cannot guarantee that such information remains private. It may also be used to train models that power AI systems.
The other is to treat anything the AI spews out as potentially false.
AI systems are still not completely secure. Tech startup Vectara estimates that AI chatbots are at least 3% likely to generate information, and in some cases as high as 27%.
A British Columbia lawyer was forced to admit in court in February that he cited two cases of family disputes fabricated by ChatGPT.
A California lawyer similarly revealed accuracy issues in April 2023 when it asked a chatbot to compile a list of jurists who had sexually harassed them. He misquoted academics and cited non-existent Washington Post articles.
Organizations creating AI policies also frequently mention the issue of transparency.
“I don't think of other people's work as my own, so why would I think of something written by ChatGPT as my own?” Canadian Institute for Advanced Research (CIFAR) Pan-Canadian Artificial Intelligence Strategy asked Executive Director Elissa Strom.
Many argue that people should be notified when it is used to parse data, write text, or create images, video, or audio, but other cases are less clear.
“You can use ChatGPT 17 times a day, but do you need to compose an email disclosing it every time? You're planning a trip and wondering whether to fly or drive, etc. Probably not,” Bhargava said.
“There are many benign cases where you don't need to disclose that you used ChatGPT.”
It's unclear how many companies have considered all the ways their employees can use AI and communicated what is and isn't acceptable.
An April 2023 survey of 4,515 Canadians conducted by consulting firm KPMG found that 70% of Canadians using generative AI said their employer had a policy regarding this technology. I found out that there is.
However, in an October 2023 survey conducted by software companies Salesforce and YouGov, 41% of 1,020 Canadians surveyed reported that their company does not have a policy for using generative AI in their work. It was concluded that it did. About 13% had only “loosely defined” guidelines.
At Sun Life Financial Inc., staff are blocked from using external AI tools in their work because the company cannot guarantee that customer, financial, and health information remains private when using external AI tools. .
However, the insurance company is allowing employees to use internal versions of Anthropic's AI chatbot Claude and GitHub Copilot, an AI-based programming assistant, only if the company adheres to both data privacy policies. said Laura Money, chief information officer.
So far, she has seen staff use the tool to write code and create notes and scripts for videos.
To encourage more experimentation, the insurer encouraged its staff to enroll in CIFAR's free, voluntary online course that teaches the principles of AI and its effects.
“We want our employees to be familiar with these technologies because it increases their productivity, improves their work life, and makes work a little more fun,” Money said of the initiative.
Approximately 400 workers have registered since the course was offered a few weeks ago.
Despite offering courses, Sun Life recognizes that AI advances so quickly that our approach to technology must continue to evolve.
For example, Plum and CIFAR announced their respective policies before generative AI tools were readily available to create sound, audio, or video beyond text.
“We didn't have the same level of image generation that we have today,” McGregor said of the summer of 2023, when Plum launched its AI policy at a hackathon, asking staff to use ChatGPT to write poems about the business and how to solve problems. He talked about how he asked people to conduct experiments to see if they could solve the problem. About business issues.
“Certainly, an annual review is probably necessary.”
Bhargava agreed, but said many organizations still don't have policies and need to catch up.
“Now is the time to do it,” he said.
“If the genie is out of the bottle, you can't think, 'Maybe next year I'll do this.'”
This report by The Canadian Press was first published May 6.
