Generative AI Is Having Its 'Herbalife' Moment: The Uncomfortable Truth

A graphic representing an economic question mark contrasting with a flashy AI technology image
AI Summary

As generative AI is introduced indiscriminately across industries, critics argue that the lack of price transparency and technical limitations are similar to multi-level marketing.

Imagine this: This morning, you commanded your AI assistant, “Summarize the meeting materials.” The AI read the document and organized the key points in an instant. But how would you feel if you received a bill later with an unimaginably high amount? Recently in Silicon Valley, somewhat provocative criticism has been emerging that generative AI is like ‘Herbalife.’ Why on earth is this being said?

Why is this important?

Generative AI (AI technology that creates new content such as text, images, and video) is rapidly changing our daily lives. Source IBM It has become possible to finish tasks that previously took hours in an instant, even with a small number of people. Source Oracle Companies are scrambling to introduce this technology into their workflows.

However, there is a point we are overlooking. It is ‘cost opacity.’ Even multi-level marketing companies like Herbalife clearly state what the initial startup costs are, but it is very difficult to know in advance how many ‘tokens’ (the basic unit of AI data processing and the standard for billing) we will consume when we ask generative AI tools to perform specific tasks. Source What We Lost

Understanding it simply: AI is a ‘statistical mimic’

First, shall we define what generative AI is? Simply put, this technology is a ‘statistical mimic.’ Source Red Hat It is a model that learns from the vast amount of data created by humanity and statistically chains together the most plausible next words or images. Source Cloudflare

By way of analogy, think of a painter who has studied an enormous amount of puzzle pieces. If we ask them to “paint the ocean,” it is similar to them statistically selecting puzzle pieces with colors and shapes similar to the ‘oceans’ they have seen before to complete the painting. The problem is that this painter does not tell us in advance how many puzzle pieces they will use until the painting is complete. Even for a single coding task, the bill amount will vary wildly depending on whether the AI solves it in one go or gives an answer after thousands of trial-and-error attempts. Source The Radical Blog

Where do we stand now?

The current tech industry is in a “try everything” heat. Source AOL Nvidia CEO Jensen Huang evaluates the current situation as a massive platform transition. Source Crescendo AI

However, there are clear limitations behind this flashy technology. Some experts point out that current AI models are nothing more than ‘probabilistic tricks,’ and that there are fundamental limitations in the very way AI is built. Source InfoWorld

A bigger problem is market concentration. Currently, the enterprise market is dominated by a few giant corporations. Source Menlo Ventures The ecosystem is so skewed that just three companies, including Google, account for a whopping 88% of enterprise AI API usage. Source Menlo Ventures

What will happen in the future?

Generative AI is set to become even smarter in the future. Moving beyond simply answering user questions, it is evolving into ‘agent’ models that set long-term goals themselves, traverse multiple apps, and handle complex tasks on their own. Source Forbes

We must enjoy the tremendous productivity that this technology brings while simultaneously contemplating how to control the opacity of the technology and its cost issues. It is a point in time where the critical perspectives of those of us who use the technology are becoming as important as the speed at which it develops.

AI’s Take: The perspective of a MindTickleBytes reporter

Generative AI is not a magic box. We must not get intoxicated by the illusion of technological innovation and accept its cost structure as a given. We must not forget that true technological innovation is completed upon a foundation of transparency. When using AI as a tool, we need the wisdom to always cautiously examine how that tool works and what we are paying in exchange.

References

  1. Generative AI Is Having Its Herbalife Moment
  2. Herbalife AI – Powered by Nowsite
  3. Generative AI is having a throw-everything-at-the-wall moment
  4. [Is the generative AI bubble about to burst? InfoWorld](https://www.infoworld.com/article/4041556/is-the-generative-ai-bubble-about-to-burst.html)
  5. Patent Landscape Report - Generative Artificial Intelligence (GenAI)
  6. CES: Generative AI Is Having Its ‘War of the Worlds’ Moment
  7. What is Generative AI? - Gen AI Explained - AWS
  8. [What is Generative AI? IBM](https://www.ibm.com/think/topics/generative-ai)
  9. [What is Generative AI? Databricks](https://www.databricks.com/discover/generative-ai)
  10. [What is Generative AI? How Does It Work? Oracle](https://www.oracle.com/kr/artificial-intelligence/generative-ai/what-is-generative-ai/)
  11. What is generative AI?
  12. What is generative AI?
  13. [What does the future hold for generative AI? MIT News](https://news.mit.edu/2025/what-does-future-hold-generative-ai-0919)
  14. The radical Blog - Generative AI Is Having Its Herbalife Moment
  15. [Global AI Adoption in 2025 – AI Economy Institute Microsoft](https://www.microsoft.com/en-us/corporate-responsibility/topics/ai-economy-institute/reports/global-ai-adoption-2025/)
  16. [2025: The State of Generative AI in the Enterprise Menlo Ventures](https://menlovc.com/perspective/2025-the-state-of-generative-ai-in-the-enterprise/)
  17. The Latest AI News + Breakthroughs in Healthcare and Medical
  18. 10 Generative AI Trends In 2026 That Will Transform Work And Life
Test Your Understanding
Q1. How do generative AI models generate responses to user queries?
  • They fully understand human intent
  • They find statistically relevant patterns from training data
  • They perform logical reasoning by thinking for themselves
Generative AI is a technology that learns from vast amounts of data and generates responses that are statistically relevant to the input information.
Q2. Why have some critics recently compared generative AI tools to 'Herbalife'?
  • Because of the innovative nature of the technology
  • Because of the lack of transparency in costs associated with token consumption
  • Because it operates as a multi-level marketing scheme
LLM (Large Language Model) tools are criticized for their cost opacity, as it is difficult to know in advance how many tokens (units of cost) will be consumed when performing a specific task.
Q3. What is the number of companies that account for 88% of enterprise LLM API usage?
  • 1
  • 3
  • 12
According to the report, three major companies account for 88% of enterprise LLM API usage.
Generative AI Is Having Its...
0:00