Menu
Our workplace experience (WX) trends series looks at recent news articles, videos, social media posts, and thought leadership pieces on workplace experience. You’ll also hear from our experts on what’s trending.
In this WX trends, we look at AI hallucinations and how to spot them. Next, are you using AI at work? These are the things you should consider. Finally, what’s being done to avoid hallucinations.
AI “hallucinations” occur when models like ChatGPT produce incorrect or misleading information, often with surprising confidence.
While tools are being updated all the time and have become less prone to errors, hallucinations still happen because of the way AI models are trained on massive amounts of data, according to a CNET article.
These models generate text that sounds plausible but isn’t always factually accurate.
Hallucinations are both a feature and a bug, the article explains. When used for creative purposes, they can be beneficial, helping to generate novel content. But for factual tasks, this becomes an issue. Tech companies, including OpenAI and Google, are working to limit these issues, but eliminating them entirely is unlikely.
Human oversight, fact-checking tools, and policies like the EU’s AI Act can help manage hallucinations, the article concludes.
“For now, users should be cautious, verifying responses and using AI models for what they excel at rather than expecting them to always provide accurate information,” says Stan Stephens, Chief Product Officer at Appspace.
Generative AI can boost efficiency, but it also comes with risks, according to an article in the Wall Street Journal. To manage these, companies are implementing guidelines.
First, watch out for bias. AI models trained on public data may reflect demographic biases, so human oversight is essential to ensure content is fair and accurate.
Second, avoid sharing sensitive business information with public AI platforms, as these may store and reuse your data, explains the article. It’s safer to use enterprise-grade AI programs designed for better security.
Third, be cautious with AI-generated content — so-called “hallucinations” can introduce false or misleading information, the article warns. Double-check sources or even train AI on your own data to reduce errors.
Transparency is crucial, especially in client-facing roles. Always disclose when content was AI-generated to avoid misrepresentation. Finally, companies should be mindful of copyright concerns, as AI might generate content that infringes on protected work.
By following these rules, businesses can tap into AI’s benefits while minimizing potential risks, the article concludes.
San Francisco startup Patronus AI, which recently secured $17 million in Series A funding, launched what it’s calling the first self-serve platform to detect and prevent AI failures in real-time.
“Think of it as a sophisticated spell-checker for AI systems, catching errors before they reach users,” says an article in VentureBeat.
“The stakes couldn’t be higher. Every time an AI system invents facts, recommends dangerous treatments, or generates copyrighted content, it erodes the trust these tools need to transform business. Without reliable guardrails, the AI revolution risks stumbling before it truly begins.”
With big-name clients like HP and AngelList already on board, and partnerships with tech giants like Nvidia and IBM, Patronus AI comes at a pivotal moment in AI development, according to the article.
“Ensuring AI systems are reliable and trustworthy will be essential for organizations moving forward,” says Stephens.
“At Appspace, we believe in the transformative power of AI, but we also recognize the importance of responsible implementation. That’s why we’ve prioritized building a platform that’s not only innovative but also secure and ensures compliance with privacy regulations and organizational policies.”
Learn about Appspace Intelligence.
© 2024 Appspace Inc. Appspace is a registered trademark of Appspace Inc. All rights reserved.