Does Not Using AI for Modeling Count as Dereliction? Musk Replied 'True,' and the Comments Split Into Two Camps

Aaron Burnett wrote on X that not using AI for modeling should be considered dereliction of duty. Elon Musk reposted it and replied with just one word: True. Aaron's follow-up was more specific: in the next 1 to 2 years, this judgment will become obvious. He compared AI modeling to Tesla FSD being used for investment decisions. You can choose to model manually, but that is not a smart decision. Every institution will have to face this fact in the coming months.

First, let's fill in the context. Modeling in investing means using spreadsheets, assumptions, and data to project returns, risks, and cash flows. FSD is assisted driving; your hands still need to be near the steering wheel. Aaron's analogy means AI should participate in the decision-making process, not that people should close their eyes and hand everything over to the model.

The opposition arrived immediately. Waterworld Capital sarcastically said this is a venture capitalist discovering they can use Claude to make up numbers in Excel. Engine Rich responded with three sarcastic lines: only a foolish carpenter uses a tape measure, only a foolish accountant uses a calculator, and only a foolish tailor uses a sewing machine. Tools are not evidence of dereliction. Austin H got closer to the core of the issue: using AI is fine; the real dereliction is trusting model output without checking assumptions. Zyrx added: models inherit biases from training data—does that count as due diligence? Claudiu said from an engineering perspective that not using available tools is already regarded as a risk. smsroute thought the FSD analogy was right, but hands need to stay near the steering wheel. Holly world asked more directly: what exactly are we agreeing to?

My judgment is that Aaron conflated tool use with judgment responsibility. Dereliction in modeling has never been about not clicking the AI button. Dereliction is failing to write assumptions clearly, failing to label data sources, and failing to review results. AI can speed up modeling, but it can also package errors more neatly. The smoother the numbers a model produces, the easier it is for people to skip follow-up questions. The FSD analogy is interesting, and risky. Driver assistance requires a person to watch the road; the steering wheel of investment decisions is assumptions, risk exposure, and explainability. Without these, AI is just a faster Excel.

The comments under this tweet were not all discussing modeling either. Some people posted good-night videos, some posted event big screens asking Musk to explain, and some posted dogs. Serious debate and noise were crammed into the same information feed.

So, does not using AI for modeling count as dereliction? It is too early to conclude. The more practical question is: after using AI, who is responsible for the assumptions, who reviews the results, and who signs off when the model goes wrong? Tools will get cheaper; responsibility will not.