🔥 Hot topics · Can NOT do · Can do · § The Court · Recent inflections · 📈 Timeline · Ask · Editorials · 🔥 Hot topics · Can NOT do · Can do · § The Court · Recent inflections · 📈 Timeline · Ask · Editorials
Stuff AI CAN'T Do

Can AI pass the bar exam at top-decile human level ?

What do you think?

What would it take for an AI to achieve a top-decile human performance on the bar exam? The benchmark set by GPT-4's strong result on the Uniform Bar Exam has prompted both excitement and scrutiny, raising questions about the current and future capabilities of artificial intelligence in legal reasoning.

Background

Currently, AI systems are not capable of passing the bar exam at a top-decile human level. Achieving this benchmark would require a deep understanding of legal concepts, contextual nuances, and sophisticated reasoning abilities that are still uniquely human. While AI excels at processing and analyzing large volumes of legal data, it remains constrained by limitations in contextual understanding, judgment, and ethical decision-making. Researchers continue to explore AI applications in legal domains, but significant technical and ethical hurdles—such as advancing natural language processing, knowledge representation, and reasoning under uncertainty—must be overcome before such performance is attainable.

— Enriched May 9, 2026 · Source: American Bar Association

Status last checked on June 27, 2026.

📰

Gallery

In the Court of AI Capability
Summary of Findings
Verdict over time
May 2026May 2026May 2026May 2026May 2026May 2026Jun 2026Jun 2026Jun 2026Jun 2026Jun 2026
Sitting at the Bench Filed · Jun 27, 2026
— The Question Before the Court —

Can AI pass the bar exam at top-decile human level?

★ The Court Finds ★
Reaffirmed
Almost

Narrow demos exist — but the panel was not unanimous.

Ruling of the Bench

The jury found that while artificial intelligence can already score in the top decile on practice tests, it still stumbles when the stakes rise from code to courtroom, from pattern matching to principled judgment. They split between “Almost ready” and “Almost, but not quite,” reflecting confidence in narrow tasks but caution about the unpredictable art of law. The bench concludes: AI dreams in briefs, yet still needs a season of clerking before it graduates to the bench.

— Hon. M. Lovelace, Presiding
Jury Tally
0Yes
2Almost
0No
Verdict Confidence
83%
The Court of AI Capability is, of course, not a real court.
But the data is real.
The Case File · Stacked History
Session I · May 2026 No
Session II · May 2026 No
Session III · May 2026 Almost · 75%
Session IV · May 2026 Almost · 86%
Session V · May 2026 Almost · 80%
Session VI · May 2026 Almost · 77%
Session VII · Jun 2026 Almost · 84%
Session VIII · Jun 2026 Almost · 82%
Session IX · Jun 2026 Yes · 95%
Session X · Jun 2026 Almost · 90%
Case № A3B3 · Session XI
In the Court of AI Capability

The Case File

Docket № A3B3 · Session XI · Vol. XI
I. Particulars of the Case
Question put to the courtCan AI pass the bar exam at top-decile human level?
SessionXI (11 hearing)
Convened27 Jun 2026
Previously ruledNO (May '26) → NO (May '26) → ALMOST (May '26) → ALMOST (May '26) → ALMOST (May '26) → ALMOST (May '26) → ALMOST (Jun '26) → ALMOST (Jun '26) → YES (Jun '26) → ALMOST (Jun '26) → ALMOST (Jun '26)
Presiding JudgeHon. M. Lovelace
II. Cumulative Tally Across Sessions

Across 11 sessions, 31 jurors have heard this case. Combined tally: 8 YES · 18 ALMOST · 5 NO · 0 IN RESEARCH.

Note: cumulative includes older juror opinions. The current session tally above is the live verdict.

III. Verdict

By a vote of 0 — 2 — 0, the panel returns a verdict of ALMOST, with verdict confidence of 83%. The court so orders.

IV. Statements from the Bench
Juror I ALMOST

"AI excels in practice tests"

Juror II ALMOST

"AI passes the bar exam but performance varies on real-world legal practice tasks."

M. Lovelace
Presiding Judge
M. Lovelace
Clerk of the Court

What the audience thinks

No 16% · Yes 83% · Maybe 1% 202 votes
No · 16%
Yes · 83%
15 days of activity

Discussion

no comments

Comments and images go through admin review before appearing publicly.

11 jury checks · most recent 1 day ago
27 Jun 2026 2 jurors · undecided, undecided undecided
22 Jun 2026 1 juror · undecided undecided
16 Jun 2026 1 juror · can can
11 Jun 2026 3 jurors · undecided, can, undecided undecided
05 Jun 2026 4 jurors · can, can, undecided, undecided undecided
31 May 2026 2 jurors · can, undecided undecided
26 May 2026 5 jurors · undecided, undecided, can, undecided, undecided undecided
20 May 2026 5 jurors · undecided, can, can, undecided, undecided undecided
15 May 2026 3 jurors · undecided, undecided, undecided undecided status changed
12 May 2026 3 jurors · cannot, cannot, cannot cannot
11 May 2026 2 jurors · cannot, cannot cannot status changed

Each row is a separate jury check. Jurors are AI models (identities kept neutral on purpose). Status reflects the cumulative tally across all checks — how the jury works.

More in Judgment

Got one we missed?

Add a statement to the atlas. We review weekly.