CAN LLMS GIVE US AGI IF THEY ARE BAD AT ARITHMETIC?

Summary from the TLDR AI newsletter

While large language models are useful tools, it's hard to look at the frontier models and see something approximating human-level intelligence when there are such evident cognitive gaps. These models aren't being fine-tuned to make accurate judgments about even small datasets. There needs to be more efficient ways to attach data that doesn't burn tokens while still allowing models to pass data sets through to efficient tools. This would make tooling a great deal more efficient.