Smale's mean value conjecture (Smale 1981) asserts that, for every complex polynomial of degree at least 2 and every point
with
,
some critical point
of
satisfies
Smale (1981) proved the weaker inequality with right side . A stronger proposed constant for degree
is
. Unlike the real mean
value theorem, the statement compares a secant quotient with a derivative
at the prescribed point and selects the other endpoint among the critical
points.
Adamczewski (2026) released a proposed counterexample to the constant 1, constructed autonomously by GPT-6 Astra. The claimed polynomial
satisfies
and
, but
at every critical
point. A Lean proof was checked against an independently
supplied formal statement with only standard logical axioms. Independent mathematical
review of the construction had not been reported as of Sep. 7, 2026 (VibeMathed
2026).
Jatar and Ng (2026) report the stronger lower bound
where
is the supremum, over complex
polynomials
of polynomial degree
at least two with
and
,
of
. They obtain
by reformulating the problem using univalent
functions and Blaschke products, and also
obtain this lower bound in every sufficiently large polynomial
degree. Jatar and Ng (2026) used GPT-6 Astra and Grok 4.6 for numerical experiments
and preliminary draft proofs, which the authors state they
verified, simplified, and rewrote.
Adamczewski and Colbrook (2026) gave a geometric construction and human exposition of counterexamples. Their explicit polynomial
in
has polynomial degree 29984, satisfies
and
, and obeys
, certified
by interval arithmetic. A small leading-term perturbation gives counterexamples in
every polynomial degree
. This supplies explicit finite-degree examples,
while the larger lower bound of Jatar and Ng (2026) concerns the value of
.