% ============================================================================
\section{Conclusion}
\label{sec:conclusion}
% ============================================================================

Using a semantic, embedding-based measure of artificial-intelligence language in the SEC current
and quarterly reports of the 500 current S\&P~500 constituents, linked to look-ahead-free forward
returns and standard factor data over fifteen formation years, we have examined whether and when
AI language was priced in the cross-section of returns. For the formation years 2010--2021 the
answer is essentially no: the top-minus-bottom quintile spread averaged $+2.5$ percentage points
per year, the six-factor alpha of the long--short portfolio was 1\% per year and insignificant, and
the annual cross-sectional slopes were of mixed sign. For the formation years 2022--2024 the
average spread was $+14.6$ points per year, the monthly long--short earned a Sharpe ratio of 1.32,
firms with high pre-launch AI language earned $+10$ points of additional abnormal return in the
year after ChatGPT, and sustainability language run through the same pipeline was penalized.

The contribution we would emphasize, however, is the anatomy of that post-2022 evidence. The
average spread rests on three cohorts and on one of them in particular: the 2024 cohort earned
$+38$ points, the 2022 cohort $+9$, and the 2023 cohort $-4$, and two pre-launch cohorts
exceeded the 2022 value. Half of the pooled post-period spread comes from a few memory and storage
manufacturers whose prices multiplied in 2025--2026. The long--short return loads heavily on a
technology-sector factor that absorbs most of its post-period alpha, sector controls halve the
cross-sectional slope and leave it insignificant, firm fixed effects absorb the effect entirely,
and a pre-determined exposure measure earned significant premia in 2015 and 2016 as well as in
2022 and 2024. The event-study difference is significant at one year under tests that allow for
cross-sectional dependence, but not at sixty days. The results are therefore consistent with a
sector-level re-rating of persistent AI exposure during the AI investment cycle, not with a stable
premium on AI language nor with a sharp attention shock to firm-specific disclosure.

Two messages emerge. The economic message is that the pricing of a disclosure characteristic can
lie dormant for a decade and then appear, and that when it appears in a short window it is easy to
mistake a sector cycle, or a single cohort, for a new factor; the discipline of cohort-by-cohort,
leave-one-out, winsorized, sector-adjusted, and dependence-robust inference is what separates the
two. The methodological message is that embedding-based measurement of disclosure content is a
productive tool, but one whose scores inherit the composition of the corpus on which they are
computed; documenting that composition, and re-estimating on homogeneous subsets of it, is part
of the method. Several next steps follow. The most important is to score the Form~10-K annual
reports, which the database has downloaded but not parsed, and to re-run the analysis on the
annual narrative that the original question concerned. The second is to reconstruct a
point-in-time index membership. The third is to add post-2024 cohorts as they become available,
which is the only way to learn whether the 2024 result was the beginning of a regime or an
episode. Beyond these, extending the panel to mid- and small-cap firms, where attention frictions
are larger, and incorporating earnings-call transcripts and patent text as complementary measures
of AI exposure, would sharpen the distinction between firm-level and sector-level repricing that
the present data cannot draw.