Billionaire investor Mark Cuban has issued a warning to the corporate world, particularly those who believe artificial intelligence will become an all-powerful entity that takes over the world. In a recent comment on social media platform X about the state of Enterprise AI, Cuban argued that the technology's greatest weakness is its lack of consistency, which makes deep industry expertise more valuable than ever.
AI's Inconsistency Problem
Cuban noted that AI models are currently incapable of providing the same answer to the same question every time. Unlike traditional software, which follows a rigid logic, Large Language Models (LLMs) produce answers based on probability theory. They essentially 'guess' the next word or action, which can lead to different results in every session, the Shark Tank star explained.
"I'm coming to the conclusion that the biggest challenge for Enterprise AI, and AI in general, as of now, is that it's still impossible to make sure that everyone gets the same answer to the same question, every time," Cuban observed. For businesses that rely on precision, this 'unreliability' is a massive liability.
A Reality Check for 'Doomers'
Cuban also used this technical limitation to address the "AI Doomers" – critics who fear that AI is on a fast track to gaining consciousness and taking over the world. He said that if an AI does not understand that its output should be consistent, it certainly does not understand the real-world consequences of what it is saying.
"[This] is a great response to the doomers. AI doesn't know the consequences of its output. Judgement and the ability to challenge AI output is becoming increasingly necessary, and valuable," he added.
Domain Knowledge Grows in Value
Cuban threw his weight behind those who have domain knowledge – deep, specialised expertise in a specific field. "Which makes domain knowledge more valuable by the second," he stated.
Tech Leaders Emphasize Human Oversight
Google CEO Sundar Pichai and Uber CEO Dara Khosrowshahi, in their companies' respective earnings calls, also backed AI-generated code but stressed the importance of human approval. Pichai revealed that Google generates about 75% of its code using AI models, but every line is reviewed and approved by human engineers.
"We've been using AI to generate code internally at Google for a while. Today, 75% of all new code at Google is now AI-generated and approved by engineers, up from 50% last fall," Pichai said in a blog post. "Every line of that code is reviewed and approved by engineers."
Khosrowshahi added that about 10% of code at Uber is written by AI but is also approved by engineers.
These insights from top tech leaders underscore a growing consensus: while AI can boost efficiency, human expertise remains indispensable for ensuring accuracy, consistency, and accountability in enterprise applications.
