Lucas Bennett @lucas_ben_ai · 24 Aug 2026
Standard AI evaluations rely on pre-existing answer keys. TRACES by @Apodex_AI asks a harder question: can AI produce a defensible result when the answer is unknown? https://t.co/BTyFJB10Bq #TRACES
28 780Views
20Likes
14Reposts
10Replies
0Quotes
0Bookmarks
Is that a lot?
4.58×vs this author's median6 290 views is typical
80Percentile for this authorof 5 recent posts
18.7×vs 10K–100K median1 540 views is typical
91.00%Reachviews ÷ followers
0.15%Engagement rateof viewers reacted