The Frankl-Ota-Tokushige conjecture concerns the number of minimum hitting sets of a uniform intersecting family.
Let
be the largest number of
-element hitting sets of a
-uniform
intersecting family whose covering number
is
. The ground set may be
any finite set. For a fixed integer
, write
as , where
denotes big-O notation.
The conjecture states
(Patkós 2026).
Patkós (2026) refuted the conjecture by proving
He also established the lower bound .
Patkós (2026) attributes the counterexample construction to ChatGPT and the
lower-bound proof to himself.
A separate conjecture of the same authors concerns the maximum size of a
-uniform intersecting family
with covering number
.
It asserts that
for some fixed
and all sufficiently large
. Frankl et al. (2026) refuted this assertion
by proving
for
. This is distinct from the fixed-
question about numbers of minimum hitting
sets. Frankl et al. (2026) attribute the new construction to generative
AI and its verification to the authors.