Coattails (adactio.com)

The piece argues for precise language: call models “large language models” (LLMs), not the catch‑all “AI.” The author says “AI” has become a marketing umbrella that conflates very different technologies—from simple rule systems to neural networks—allowing actors to borrow credibility across domains. That conflation lets successes in areas like medical ML or computer vision be waved as proof for LLM claims, and opens the door for hype‑driven “grifters” to overstate LLM capabilities. The essay even points to a rhetorical shift where reputable services now label themselves “traditional machine learning,” a deliberate distancing from the LLM/“AI” brand. This matters to the AI/ML community because naming shapes public understanding, policy, funding and trust. Technically, LLMs differ from many “traditional” ML systems in scale, architecture, training data (web‑scale unsupervised/causal language modeling vs task‑specific supervised learning), evaluation metrics, interpretability and failure modes. Treating them as interchangeable hides these differences and risks tarnishing robust, domain‑specific methods if an LLM bubble bursts. The observed rebranding could therefore protect established ML and computer vision work from reputational contagion, influence regulation and procurement choices, and clarify research and deployment expectations across subfields.