In A World of AI Noise, What Can You Trust? Six Dos and Don’ts

The LinkedIn dashboard is only ever a few clicks away, so at any time, any of us can find as many hot takes on AI as we'd like—everything from "AI will replace every worker" to "AI changes nothing" to "Here's how I used ChatGPT to revolutionize my skincare routine." You need to know how to tell a real AI trend from one that's designed to get headlines.  So how do you know what to trust? 

The best advice would be to use your own critical thinking to make your own judgments about what makes sense—and that applies whether you’re making decisions either for yourself or for your organization’s labor market strategy. Here are six ideas about what to trust and what to be skeptical of. 

And why should you trust us? That’s the right question to ask. In addition to the reputable sources you’ll find linked throughout here, Lightcast has over 25 years of experience analyzing millions of actual job postings, salary data, and workforce patterns—not speculation or wishful thinking. More specifically, many of these ideas come up in “Beyond the Buzz: Developing the AI Skills Employers Actually Need,” our latest research. 

Don't Trust: LLMs Themselves

Yes, ChatGPT can write a decent email. Claude can help you brainstorm. But large language models aren’t search engines; they craft responses based on words that sound good together, and they can’t evaluate the accuracy of what they say. So maybe that's fine if you're trying to come up with another way to phrase a sentence, but you shouldn’t take their responses as a source of truth. The only exception might be AI tools that provide footnotes, like Perplexity or Google AI Overview, but even then, it’s up to the user to decide if those sources are reliable. 

Trust: Prompts Shared By Other Humans

When someone shares how a specific AI tool helped them solve a particular problem, there’s value in that insight. The copywriter who cuts research time in half using AI-powered competitive analysis isn't making a grand prediction about the future of work—they're sharing a practical workflow improvement. These ground-level insights can’t be discounted out of hand. 

For example, here’s a free tip: chatbots respond better to questions that use good manners, your classic “please” and “thank you.” That’s because they mimic human behaviors that show up in text, and humans respond better to kind questions and instructions than brusque ones. 

Don't Trust: Job Postings To Tell You How Many Workers Use AI

Millions of workers are using AI in jobs they were hired for before 2022—before most companies even knew AI would jump beyond tech. Lightcast data tells us that over 80,000 job postings in 2024 mentioned generative AI skills, but that’s not to say only 80,000 workers were using generative AI at work. The absence of AI requirements in existing job postings doesn't mean AI isn't being used; it means that job postings tell us something different about the labor market.

Trust: Job Postings To Tell You Which AI Trends Employers Are Following

While job postings lag behind current usage, they're excellent predictors of where employers are headed. So when 80,000 job postings in 2024 mentioned generative AI skills, that doesn’t tell us how many workers currently use AI, but it does provide reliable evidence of what companies are prioritizing as they build their teams. 

The market doesn't lie about what it values. When job postings mentioning AI skills advertise salaries 28% higher than those that don't—roughly $18,000 more per year—that's not hype. That's measurable value. The biggest jump happens with the first AI skill, meaning organizations can see immediate returns on targeted training investments.

Don't Trust: A One-Size-Fits-All Approach

"Every employee needs AI training" sounds progressive, but it's strategically unclear. The AI skills a marketing manager needs (generative AI, natural language processing) are quite different from what a maintenance technician requires (robotics, visual recognition). Generic "AI literacy" programs often produce well-meaning generalists without practical application.

When someone makes sweeping statements about "the future of work," be curious about their methodology. Predictions that sound authoritative but lack specific data sources are often more marketing than analysis.

Trust: A Customized Strategy for Your Specific Needs

Real insights come from understanding that different fields are adopting AI at different speeds and in different ways. Marketing and PR roles show 50% growth in AI requirements, while transportation is taking a different path. Science and research positions emphasize machine learning expertise, while customer service roles focus on natural language processing capabilities. These aren't abstract trends—they provide specific intelligence about where to invest training resources

Cutting Through The Noise

Our idea with the Beyond the Buzz report was to avoid baseless speculation around AI and instead provide hard data from millions of job postings. With Lightcast data, we’re able to identify which specific AI skills employers actually pay premiums for, which roles require them most urgently, and how different roles are adapting at dramatically different speeds. AI has already rewritten the rules of workforce value, and Beyond The Buzz tells you how.

Thanks for reading On The Job. Be sure to catch up on our past issues, including "Return-To-Office is Creating Two Classes of Worker," and "The Next Great Resignation,“ and you can also subscribe here. We’ll see you next time.