Can Machines Think? A Resolution of the Lovelace-Turing Debate

Can Machines Think?

Revised concise version

This is a shorter revision of the manuscript. It starts from section 2.6 of the Stanford Encyclopedia of Philosophy’s Turing Test entry. The core claim is unchanged: the right target is originativity, not ex nihilo creation.

Machines will produce correct proofs, boring proofs, and wrong proofs. That does not collapse research into proof checking. The scarce skill moves upstream to problem selection, definition choice, explanatory framing, and judgment.

Full Manuscript

Read the concise preprint here:

• GitHub Gist: can machines think?

Why this matters

• Verification is easier than search, but research is not just search.
• AI will lower the cost of generating candidate proofs and arguments.
• The human task becomes deciding what is worth proving and why it matters.
• That judgment is the research skill the paper defends.

Citation

Om. 2026. Can Machines Think? A Resolution of the Lovelace-Turing Debate (concise revised version). Preprint. Available via GitHub Gist.

If you want to share one link, share this page. If you want the manuscript itself, share the gist above.