The Rise of AI Washing

‍

It’s AI—sorta.

‍

It’s costly to build a service powered by artificial intelligence. You could build an application on existing large language models (LLMs) like OpenAI’s ChatGPT for $20,000-150,000, but creating an LLM using proprietary data will cost millions on the low end. Databricks, a cloud-based data platform, just spent $10 million on its generative AI model.

‍

Since many companies don’t have the capital or resources to build their own AI models, they are paying humans, often in other countries for low wages, to behave like robots so they can sell “AI services” to customers and their investors. 

‍

Some examples of AI washing include:

‍

• Amazon’s “Just Walk Out” relied on more than 1,000 people in India watching and labeling videos, per Gizmodo
• Arcads platform to “create winning ads with AI Actors” is a collection of human actors, not AI actors, performing the work behind the scenes
• X.Ai had humans pretend to be chatbots for 12 hours per day to create its “AI calendar scheduling services,” per Bloomberg

‍

The SEC is starting to take notice. In March, the SEC announced charges against two investment advisors, Delphia (USA) Inc. and Global Predictions Inc., for making false and misleading statements about their purported use of AI.

‍

While the implications of lying to customers, investors, and partners about the use of AI should not be understated, the situation offers some low-hanging fruit for joke material–one Twitter user called AGI (artificial general intelligence) = a guy in India.