{"id":1993,"slug":"gpt-3","name":"GPT-3","type":"concept","created_at":"2026-08-08T04:45:06.226392+00:00","hidden":0,"aliases":[],"episodes":[{"id":"e6f3567c-989d-4b50-bbea-b89bff7d51cf","slug":"s14e08-some-combination-of-efficiency-and","subject":"s14e08 - Some Combination of Efficiency and Expediency","title":"Some Combination of Efficiency and Expediency","publish_date":"2023-01-17T17:00:34.382101+00:00","year":2023,"month":1,"season":14,"episode":8,"episode_label":"s14e08","canonical_url":"https://newsletter.danhon.com/archive/s14e08-some-combination-of-efficiency-and/","is_premium":0,"word_count":393,"context":" … On the other hand, Simon Willison continues to inspire by creating semantic search by what I can only inexpertly describe as\nexpertly mashing up Datasette and the <mark>GPT</mark> … ","reading_minutes":2},{"id":"c2063e91-324e-4ed0-92be-fc388eac3623","slug":"s13e16-duolingo-but-for-talking-to-ml-image","subject":"s13e16: Duolingo, but for talking to ML Image Generators","title":"Duolingo, but for talking to ML Image Generators","publish_date":"2022-10-27T16:23:01.112416+00:00","year":2022,"month":10,"season":13,"episode":16,"episode_label":"s13e16","canonical_url":"https://newsletter.danhon.com/archive/s13e16-duolingo-but-for-talking-to-ml-image/","is_premium":0,"word_count":1316,"context":" … Which brings me to Duolingo, but for learning the language of machine learning prompts, whether they're image generation or conversational \"finish this sentence\" <mark>GPT-3</mark>/text generation … ","reading_minutes":7},{"id":"20341722-0a5e-4cfb-84ba-601dcc36a463","slug":"s12e45-bits-and-pieces","subject":"s12e45: Bits and pieces","title":"Bits and pieces","publish_date":"2022-08-04T23:53:03.171239+00:00","year":2022,"month":8,"season":12,"episode":45,"episode_label":"s12e45","canonical_url":"https://newsletter.danhon.com/archive/s12e45-bits-and-pieces/","is_premium":0,"word_count":830,"context":" … DALL-E and so on generating slideware/presentation decks (today, prompted by the wonderful HBO/Discovery earnings call), <mark>GPT-3</mark> being prompted into generating malicious payloads, and finally … ","reading_minutes":4},{"id":"bd550ce7-121d-4212-92ce-e6740a9291d6","slug":"s12e27-just-put-dr-ian-malcolm-on-our-tombstone","subject":"s12e27: Just Put Dr. Ian Malcolm On Our Tombstone Already","title":"Just Put Dr. Ian Malcolm On Our Tombstone Already","publish_date":"2022-06-23T23:35:55.414513+00:00","year":2022,"month":6,"season":12,"episode":27,"episode_label":"s12e27","canonical_url":"https://newsletter.danhon.com/archive/s12e27-just-put-dr-ian-malcolm-on-our-tombstone/","is_premium":0,"word_count":1439,"context":" … come up with a bunch of language models like\n<mark>GPT-3</mark>\nthat people get to use (over 300, as of March 25, 2021)\n\ncome up with a bunch … ","reading_minutes":7},{"id":"849e102d-42f5-4364-890c-ccef786ef254","slug":"s08extra01-the-one-about-writing","subject":"s08extra01: The One About Writing","title":"The One About Writing","publish_date":"2020-08-06T14:19:16.294000+00:00","year":2020,"month":8,"season":8,"episode":1,"episode_label":"s08extra01","canonical_url":"https://newsletter.danhon.com/archive/s08extra01-the-one-about-writing/","is_premium":1,"word_count":1130,"context":" … <mark>GPT-3</mark>’s unique point of view is “nothing has ever ingested the amount of textual content in this way as <mark>GPT-3</mark> has”, and thus nothing is … ","reading_minutes":6}]}