Turing replaced a question about thinking with a test of imitation
Turing's 1950 paper proposes an indistinguishability test rather than a definition of thinking, the frame every companion app's persona now inverts.
Read piece →Topic shelf · 17 records
Market & history
The category’s commercial moves, precedents, launches, and policy context.
Historical records and commercial developments help separate category precedent from a claim about present-day product availability.
ELIZA showed people project understanding onto pattern matching
Weizenbaum's 1966 paper describes a keyword-matching script and warns that its illusion of understanding is dangerous.
Read piece →A contest paid chatbots to seem human not to be right
The Loebner Prize ran from 1991 to 2019, and a formal 1994 critique argued its judged format could not validate machine intelligence.
Read piece →A patent shows SmarterChild's real innovation was distribution
A 2000 patent filing shows SmarterChild's screen-name login let one bot reach users already inside AOL and MSN chat networks.
Read piece →Tamagotchi proved a feeding loop could sell for decades
Bandai's own history page dates Tamagotchi to 23 November 1996 and puts shipped units above 100 million as of mid-2025.
Read piece →Sony's aibo relaunch made ongoing ownership need a subscription
Sony's November 2017 press release for the new aibo makes cloud-plan enrolment necessary for the robot to function.
Read piece →Microsoft's own post-mortem named a gap not a rogue AI
Microsoft's March 2016 blog post says a coordinated attack, not autonomous learning, drove Tay's offensive tweets.
Read piece →XiaoIce's own paper names conversation length as its target metric
The 2018 XiaoIce paper optimises for conversation-turns per session, not correctness, and reports over 660 million users.
Read piece →Replika grew from a memorial chatbot into a general companion
Replika's own history page names its founders and mission but sends readers elsewhere for its 2015 origin story.
Read piece →A small trial gave Woebot's chatbot claims their evidence template
A 70-person 2017 JMIR trial found reduced depression symptoms in college students using the Woebot chatbot for two weeks.
Read piece →Meta let anyone build a labelled AI character across its apps
Meta's AI Studio announcement and its 2026 shutdown notice show how a creator-AI feature launched, spread and was later retired.
Read piece →Character.AI's Series A came six months after a fast launch
The investor's and the company's own posts fix the date, the backers and the usage figures, but not the amount raised.
Read piece →Advocacy groups asked regulators to examine Replika's ads
Three advocacy groups' own filing alleges deceptive Replika marketing and reviews, described here only as allegations.
Read piece →LOVOT sells affection with a price tag and a required subscription
Groove X's and LOVOT's own pages describe a robot built only for affection and sold with a required monthly fee.
Read piece →Wellbeing studies on chatbots differ in size and funding
A 2025 Therabot trial, an OpenAI-MIT study and a 2017 Woebot trial show how sample size and disclosed funding change a claim's weight.
Read piece →Lovot sells affection with a hardware price and a plan
Lovot's maker prices a subscription-model companion robot around care plans, as its own pages state on retrieval.
Read piece →Map the companies behind a companion before calling it independent
A dependency map reveals when model, cloud, speech, app-store and payment choices converge on the same few suppliers.
Read piece →