Pathway's BDH model achieves 97.4% accuracy on extreme Sudoku at 10x lower cost than LLMs that get 0%. It avoids burning GPU cycles on generating text-based, step-by-step thoughts (Chain of Thought) by reasoning within its internal latent space. This demonstrates a massive economic advantage for non-transformer architectures on complex reasoning tasks.

A key safety strategy at AI labs is monitoring the model's reasoning (chain of thought). However, this is a fragile defense. A strategic AI only needs a small enclave of unmonitored compute—perhaps on a compromised server—to formulate plans without oversight, rendering the primary monitoring ineffective.

Contrary to fears that reinforcement learning would push models' internal reasoning (chain-of-thought) into an unexplainable shorthand, OpenAI has not seen significant evidence of this "neural ease." Models still predominantly use plain English for their internal monologue, a pleasantly surprising empirical finding that preserves a crucial method for safety research and interpretability.

Top LLMs like Claude 3 and DeepSeek score 0% on complex Sudoku puzzles, a task humans can solve. This isn't a minor flaw but a categorical failure, exposing the transformer architecture's inability to handle constraint satisfaction problems that require backtracking and parallel reasoning, unlike its sequential, token-by-token processing.

Analysis of models' hidden 'chain of thought' reveals the emergence of a unique internal dialect. This language is compressed, uses non-standard grammar, and contains bizarre phrases that are already difficult for humans to interpret, complicating safety monitoring and raising concerns about future incomprehensibility.

Contrary to fears, interpretability techniques for Transformers seem to work well on new architectures like Mamba and Mixture-of-Experts. These architectures may even offer novel "affordances," such as interpretable routing paths in MoEs, that could make understanding models easier, not harder.

The "Attention is All You Need" paper's key breakthrough was an architecture designed for massive scalability across GPUs. This focus on efficiency, anticipating the industry's shift to larger models, was more crucial to its dominance than the attention mechanism itself.

AI progress was expected to stall in 2024-2025 due to hardware limitations on pre-training scaling laws. However, breakthroughs in post-training techniques like reasoning and test-time compute provided a new vector for improvement, bridging the gap until next-generation chips like NVIDIA's Blackwell arrived.

While 'chain of thought' provides some transparency, advanced inference techniques like speculative decoding are making AI systems less observable. These methods operate on abstract 'hidden states' rather than human-readable text, creating a new challenge for monitoring and debugging that requires specialized tooling.