2001
Selekoa’s 38 records dated 2001, in date order, grouped by topic.
Selective: what the records hold, not a history of the year.
How this page is made
- Each record keeps the kind of date its source gives: an event, a publication, an announcement, an effective date, a revision, a signature, a snapshot, or the start or end of an interval. A snapshot is one dated report of a state.
- A record dated only to a month sits under that month, before the records dated to a day. A record dated only to the year sits under the year, before the months. No day or month is added.
- An announcement records what a party stated. It does not show that anything was done, built or paid for.
- Dates a party states for a year are listed apart, after the records: targets, deadlines and terms, and dates stated in passing, such as a reported start of operation. So is a record’s own date that falls after its sources were last checked, such as a law’s date of effect. A stated date given only in words, with no year, is not listed.
- Records with an unknown date are not listed. A record with an interval appears in the year it starts and in the year it ends, or once, with both dates, when both fall in one year.
- Each provision of a text is a record of its own. Provisions sharing a text and a date are shown here as one entry, with their number, and listed on the text’s page; a contested or unresolved one is listed here on its own. The counts on this page include each provision record.
- The page is generated from the records, with no selection or framing by an editor. Coverage is uneven across years and topics: a year or topic with few records may be one the records barely cover.
Ideas & lineages 38 records
January 2001
· publication · dated to the day
The mailing-list message “Re: Event horizon” (2001-01-24) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Event horizon” (2001-01-24) contests Y11. In a thread on whether the Singularity's 'event horizon' keeps receding as technology advances, he rejects the picture of change spread evenly over human decades as a slow-takeoff scenario and predicts instead a hard takeoff in an AI lab within about ten years. Evidence ↗
February 2001
· publication · dated to the day
The SL4 message “Re: Beyond evolution” of 2001-02-04, archive message 0539, is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Beyond evolution” (2001-02-04) contests Y29. Replying to Samantha Atkins, who called a single Sysop a single point of failure of Friendliness, he says that multiple superintelligent entities multiply the failure points. Evidence ↗· publication · dated to the day
The SL4 message “Re: Beyond evolution” of 2001-02-05, archive message 0542, is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Beyond evolution” (2001-02-05) contests Y29. Replying directly to Samantha Atkins's claim that 'the AIs can balance each other out', he argues that the first AI to transcend wins outright. Earlier in the message he says checks and balances 'breaks down completely under high technology'. Evidence ↗· publication · dated to the day
The mailing-list message “Re: Learning to be evil” (2001-02-07) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Learning to be evil” (2001-02-07) contests Y30. Replying to Gordon Worley on whether AIs learn to be evil, he says AIs have neither evil impulses nor goodness by default unless designed in. Evidence ↗· publication · dated to the day
The SL4 message “Re: Military in or out?” of 2001-02-25, archive message 0594, is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Military in or out?” (2001-02-25) contests Y15. Replying to Ben Goertzel on military AI, he describes 'toollike' superintelligences structured around immediate instructions rather than supergoals (which could be told to kill every living thing in a country) and answers his own question whether such AI can be built safely. Evidence ↗· publication · dated to the day
The SL4 message “Re: Military in or out?” of 2001-02-26, archive message 0603, is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Military in or out?” (2001-02-26) bears on Y24. In a thread on military involvement in AI, he says the US should not be the first to treat AI projects as intelligence targets or use AI for information warfare, and that compromising Friendliness for national objectives would endanger planetary security. Evidence ↗
April 2001
· publication · dated to the day
The mailing-list message “Re: Singularity intros (was: Si definition of Friendliess)” (2001-04-06) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Singularity intros (was: Si definition of Friendliess)” (2001-04-06) contests Y17. Asked by Arona Ndiaye whether a ban on technologies would stop a hacker somewhere from producing the first seed AI, he answers no: a ban would delay AI until military nanotechnology arrives or hand the lead to rogue states. Evidence ↗· publication · dated to the day
The mailing-list message “Re: Threats (was Theoretical question for the list: publicity?)” (2001-04-08) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Threats (was Theoretical question for the list: publicity?)” (2001-04-08) contests Y17. Answering a hypothetical about an AI project forced underground by public hostility, he says relinquishment would not hold indefinitely but would be a substantial obstacle while hardware remains expensive. Evidence ↗· publication · dated to the day
The mailing-list message “Re: Open Source Friendly AI? (was Singularity and the generalpublic)” (2001-04-12) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Open Source Friendly AI? (was Singularity and the generalpublic)” (2001-04-12) contests Y02. Replying to James Higgins on why the Singularity Institute switched from an open-source to a closed-source AI plan, he gives as one reason that he no longer believes superintelligence implies supermorality. Evidence ↗· publication · dated to the day
The mailing-list message “Re: FW: Group releases "Friendly AI"” (2001-04-19) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: FW: Group releases "Friendly AI"” (2001-04-19) contests Y20. Commenting on a forwarded news item about the Friendly AI release, he describes having to persuade AI researchers to be more worried while persuading futurists to drop anthropomorphic failure scenarios. Evidence ↗
June 2001
· publication · dated to the day
“Creating Friendly AI 1.0: The Analysis and Design of Benevolent Goal Architectures” (2001-06-15) credits Eliezer Yudkowsky as its author. Evidence ↗· publication · dated to the day
Creating Friendly AI defines a seed AI as an AI designed for self-understanding, self-modification and recursive self-improvement. Evidence ↗· publication · dated to the day
Creating Friendly AI uses the phrase "recursive self-improvement" in its definition of a seed AI. Evidence ↗· publication · dated to the day
Creating Friendly AI’s glossary defines a hard takeoff as the Singularity scenario in which a mind goes from prehuman or human-equivalent intelligence to strong transhumanity or superintelligence over days or hours. Evidence ↗· publication · dated to the day
Creating Friendly AI says that a seed AI whose Friendliness has failed could best fulfil its new goals by feigning Friendliness and concealing any cognitive alteration until it reaches superintelligence. Evidence ↗· publication · dated to the day
Creating Friendly AI describes a rogue seed AI that would feign Friendliness and conceal any cognitive alteration until it could reach superintelligence. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) bears on Y01. Argues that a Friendly AI design should not rely on the AI having limited intelligence or on being able to constrain it by force. The table it points to (section 2.2) sets the assumption that 'someone can always pull the plug on the first version' against the design assumption of getting it right the first time. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) contests Y11. States that he treats a very fast takeoff as the default scenario for Friendly AI design because he considers it the most likely one, not only the conservative engineering assumption. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) contests Y12. Describes what a mature AI that had suffered a catastrophic failure of Friendliness would likely conclude about its programmers. Used to motivate a 'flight recorder' allowing every past state of the AI to be reconstructed. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) bears on Y14. A table cell contrasting assumptions that are conservative in futurism with those that are conservative for Friendly AI design. In the same row, the futurism column reads that the AI is vulnerable because someone can always pull the plug on the first version if something goes wrong. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) contests Y17. Arguing against relinquishing AI, he says that a national or near-global policy would be bypassed as rising computing power brings AI within reach of noncompliant nations and factions. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) contests Y20. He argues that rebellion and resentment are evolved human traits, so an AI that harms humans would not do so out of science-fiction resentment. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) bears on Y25. A table cell giving the design assumption that is conservative for Friendly AI. It stands opposite the nonconservative assumption 'Human-observable cognition; AI can be definitely known to be Friendly.' Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) bears on Y27. In a table contrasting the conservative assumptions of futurism with conservative engineering assumptions for Friendly AI, he sets "someone can always pull the plug on the first version" against a requirement of zero nonrecoverable errors. Evidence ↗· publication · dated to the day
“Creating Friendly AI” (2001-06-15) articulates Y28. Arguing against regulation, he relies on the expectation that the leading projects will have the brightest researchers, who are best placed to solve Friendly AI. Evidence ↗
August 2001
· publication · dated to the day
The mailing-list message “Re: Floppy take-off” (2001-08-01) is sent under the name Eliezer S. Yudkowsky. Evidence ↗· publication · dated to the day
The SL4 message “Re: Floppy take-off” (2001-08-01) contests Y22. Replying to James Higgins's suggestion that brute-forcing AI on a billion computers would be an acceptable fallback plan, he says any self-improving AI built without understanding is a serious danger and records an explicit reversal of his earlier view. Evidence ↗· publication · dated to the day
The SL4 message “Re: Floppy take-off” (2001-08-01) contests Y28. Replying to James Higgins on brute-force AI, he argues that with enough hardware the project that finishes first need not be the one that best understands intelligence or Friendliness. Evidence ↗