Recognizing that LLMs principally function via autocomplete could explain Sam Altman’s defensive response. Trillions of spending may not realize the AGI revolution he envisioned.

For more, read our blog post: buildaligned.ai/blog/a-lette...

In simply leveraging pre-existing data to follow prevailing forms of thought, LLMs have struggled to make the novel scientific discoveries and research advancements that many have propounded.

Yet as of now, LLMs operate according to next word prediction. They use vast troves of data to recognize linguistic patterns and determine a statistically feasible response. Lacking a mental lexicon to draw from, they fail to interpret meaning behind the words they 'know.'

As our Rebecca Gorman articulates in A Letter on the State of AI, Open AI relies on ostensibly infinite capital to train its models. The hope is that with enough compute data, they will develop a model that can achieve the elusive ‘emergent consciousness’ propounded by AGI enthusiasts.

arxiv.org/pdf/2508.15748

buildaligned.ai/blog/ai-chap...