As AI technology spreads across industries, so do the hidden risks.

AI errors could have serious consequences.

My ad agency uses AI in multiple ways… all with caution. I use AI and Chat GPT for multiple purposes, including research and ideas.

However, AI experts and I have discovered that Artificial Intelligence generates more false information than many realize.

These errors are often called “hallucinations.”

For example, with Chat GPT, my bio has 12 factual errors, but it can be worse than just factual errors.

Take OpenAI’s tool, Whisper, for example.

Whisper is designed to convert spoken language into written text, functioning as an automatic speech recognition (ASR) system.

But it has a major flaw: it often “hallucinates” or invents sentences and even entire passages. Numerous software engineers, developers, and researchers confirm that Whisper frequently produces these fictitious transcriptions.

The content of these hallucinations is especially concerning, sometimes including racially charged language, violent themes, or even fabricated medical information.

Experts who analyzed more than 100 hours of Whisper’s transcriptions found hallucinations in roughly half of them, with some developers claiming hallucinations appear in nearly every transcript they produce.

This issue is even more alarming because Whisper is used for closed captioning, a critical tool for the deaf and hard-of-hearing community.

These users are particularly vulnerable to misinformation in transcriptions, as they cannot distinguish real content from fabricated text buried within their captions. This is why research and copy are becoming less reliable, not more.

Christian Vogler, director of the Technology Access Program at Gallaudet University, has voiced concerns about the potential harm this could cause, warning that such “hidden” fabrications are hard to detect.

This issue isn’t confined to low-quality recordings. A recent study analyzed thousands of well-recorded audio samples and discovered hundreds of hallucinations.

Experts caution that these errors could proliferate across various industries, as Whisper is widely used to transcribe interviews, generate written text, and provide subtitles on popular platforms.

Remember, one of the reasons AI struggles is that it uses existing information, including AI past searches, and it builds upon it as fact.

The consequences of these faulty transcriptions are real, especially when applied in sensitive areas like healthcare, where the accuracy of medical records is paramount.

Despite warnings that Whisper should not be used in high-stakes environments, many hospitals rely on it to transcribe patient consultations.

Alondra Nelson, former head of the White House Office of Science and Technology Policy, highlights the dangers of relying on faulty transcriptions in medical settings, saying, “Nobody wants a misdiagnosis.” According to Nelson, AI tools used in critical environments must meet a “higher bar.”

With roughly 4.2 million downloads each month, Whisper’s reach is immense. Sanchit Gandhi, a machine-learning engineer, describes Whisper as the most popular open-source speech recognition model, integrated into everything from call centers to consumer voice assistants.

Given the volume of its usage, the risk of widespread misinformation in AI-generated transcriptions is higher than ever.

AI tools like Whisper promise groundbreaking advancements but also have serious pitfalls.

As these systems weave into the fabric of healthcare, media, and everyday communication, their “hallucinations” could lead to real-world harm—from misdiagnosed patients to misinformed consumers.

While AI offers remarkable potential, it must be developed and deployed with stringent accuracy checks, especially in high-stakes environments.

Until then, relying on these tools without careful, responsible government oversight must be done with due diligence and caution.

 

Unlock Explosive Growth with CDMG, Inc.
Your Partner for Digital Success

In today’s fast-paced world, you need more than just an agency—you need a partner who drives growth, amplifies your brand, and delivers results that matter. That’s where CDMG comes in.

We’re not just another ad agency. We are a top advertising and marketing powerhouse that specializes in driving real, measurable results and profits. Our award-winning team, led by marketing veterans Craig and Caleb Huey, is dedicated to supercharging your advertising efforts and transforming your business.

Why CDMG?

  • Proven Success: 120 prestigious awards in 2024 alone.
  • Data-Driven Strategy: We don’t guess. We know what works. Our campaigns are crafted with precision to deliver maximum ROI.
  • Expertise Across the Board: From media buying to digital copywriting, testing, and direct-to-consumer strategies—we’ve got it all covered.
  • Real Results: We’ve helped launch new products, reposition brands, and turn small businesses into multimillion-dollar success stories.

Whether you’re looking to launch a new product, reposition your brand, or supercharge your e-commerce growth, CDMG is the partner you need to dominate your market.

Are you ready to see how we can drive your growth?
Let’s talk. Call Michael Oppenheimer at 615-933-4647 or email [email protected] today. Don’t wait—start transforming your marketing results now with CDMG, Inc.

Act now, grow faster, and achieve more with CDMG.