Nick Heer, writing on his blog Pixel Envy, analyzes a recent Wired article by AI critic Gary Marcus questioning the viability of generative AI. Marcus argues that generative AI systems are fundamentally flawed, describing them as “autocomplete on steroids” that cannot fact-check their own output. The critic points to OpenAI’s reported financial challenges and customer disappointment with ChatGPT to support his skepticism about the technology’s future.
However, Heer challenges Marcus’s reasoning, noting that the financial figures cited were inaccurate and that OpenAI recently raised $6.6 billion at a $157 billion valuation. Heer emphasizes that generative AI extends beyond chatbots, highlighting successful applications in creative tools like Adobe Lightroom Classic’s Generative Remove feature, code development, and video enhancement. He points out that while Marcus focuses heavily on OpenAI’s business challenges, these don’t necessarily reflect on the broader utility of generative AI technology. Heer suggests that rather than a complete failure of the technology, the industry might be moving toward commoditization, with established companies like Adobe, Google, and Microsoft continuing to integrate AI features into their products.