AI Futures Project
AI Futures Project coauthor, stating his own median · Apr 2, 2026
Source published Apr 2, 2026
person
Daniel Kokotajlo
Daniel Kokotajlo develops AI forecasts and scenarios, including AI 2027; scenario assumptions and revisions remain distinct from observed outcomes.
Escape Meridian selects and paraphrases these records. A mention, shared topic or affiliation does not establish authorship, endorsement, agreement or causation. Missing positions remain unknown.
Current readings
Publisher connections and mentions are labelled separately from a person’s own statements.
Checked all 12 current publications. Showing up to 30 connected readings, ordered by original source date.
No current reading is connected here. Explore the dated statements and source references below.
Background & sources
AI forecaster and AI 2027 coauthor
Produces explicit AI scenarios and engages directly with disputed learning timelines.
Scope and interests
AI progress and research automation; scenario writing is not an observed outcome.
Coauthors AI 2027; his published forecasts are part of the argument being examined.
Profile reviewed Oct 14, 2026. Historical roles stay attached to their source dates.
Identity references
Roles & affiliations
These links preserve sourced role context. A person’s statements are not automatically the institution’s position.
Dated statements
Newest source date first. The date describes the source; it does not claim the view is current.
Automated coding staff by mid-2028
Kokotajlo's median for the Automated Coder milestone moved from late 2029 to mid-2028. The milestone is the point at which an AGI company would rather lay off all its human software engineers than stop using AI for software engineering.
AI Futures Project coauthor, stating his own median
Original source · AI Futures ProjectAttribution, date and limits
Byline date reads Apr 02, 2026. Opening paragraphs stating Daniel's Automated Coder median and defining the milestone
Post text retrieved. The median sentence is specifically Daniel's. Eli Lifland's median is stated separately in the same post.
The milestone is replacement of human software engineers at an AGI company, not learning across all white-collar work. The post attributes the shorter median partly to a faster coding-time-horizon trend and to lowering his time-horizon requirement after Opus 4.6.
Late 2029 is the previous median as reported in this post; the earlier source of that figure was not checked in this review. Mid-2028 is not converted into a month. A median is recorded as 50% because he called it a median.
Automated Coder is not the end-2028 humanlike-learning milestone in his June 2025 reply, and not an experienced-editor forecast.
Editorial record as of Oct 14, 2026.
AI research could shorten the wait
Kokotajlo gives end-2028 as his median for the learning milestone, anticipating faster progress after automated coding and research.
AI forecaster responding to Patel
Original source · Dwarkesh PodcastAttribution, date and limits
Date of this author's comment, not the parent essay's date. Daniel Kokotajlo's dated comment below the essay
Original text reviewed; audio/video not independently verified. The link may reflect publisher edits after publication.
Faster AI research is a central assumption.
A forecast in a direct reply, not a demonstrated learning system.
His reply explicitly identifies Patel's continual-learning milestone. It is not an independent replication of Patel's observations.
Editorial record as of Oct 14, 2026.
Comparisons & changes
Connections concern particular claims and dates. Sequence alone does not show influence or causation.
No sourced comparison or change is recorded here. Shared topics and affiliations do not fill that gap.
Source library
Daniel Kokotajlo — existing tracker reference
Connect the existing canonical profile to its evidence in the source corpus without adding a new statement or relationship registry.
Context and limitations
A cited biography, joint paper, interview or organization page is identity/context evidence, not a personal feed or proof of current posting; institutional views do not transfer to this person.
Public reference already used by the sourced tracker profile; access and publication-channel readiness are assessed separately.