corpus.blog
41554 blogs · [ { "id": "01a0b0c7-e26f-72fe-8617-f8420d518aae", "title": "Localising NLG content in a driving feedback system", "url": "https://ehudreiter.com/2026/09/17/localising-nlg-content-in-a-driving-feedback-system/", "published_at": "2026-09-17T07:55:32+00:00" }, { "id": "01a08216-4cc8-73dd-ad18-047d9fa151f7", "title": "Desk rejection is an unfortunate necessity", "url": "https://ehudreiter.com/2026/09/08/desk-rejection-is-an-unfortunate-necessity/", "published_at": "2026-09-08T08:26:50+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee5462a8f4b", "title": "Cycling in Northern Ireland", "url": "https://ehudreiter.com/2026/08/25/cycling-in-northern-ireland/", "published_at": "2026-08-25T06:52:48+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee546bdf00b", "title": "AI for Healthcare: Great benchmark scores but minimal impact", "url": "https://ehudreiter.com/2026/08/05/aimed-great-benchmarks-minimal-impact/", "published_at": "2026-08-05T08:29:55+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee5477fe2af", "title": "Memories of being an NL researcher in 1990", "url": "https://ehudreiter.com/2026/07/20/memories-of-being-an-nl-researcher-in-1990/", "published_at": "2026-07-20T08:32:14+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee547873581", "title": "What is the purpose of ACL conferences?", "url": "https://ehudreiter.com/2026/07/09/what-is-the-purpose-of-acl-conferences/", "published_at": "2026-07-09T08:33:49+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee547bf2a5e", "title": "Future of NLG evaluation", "url": "https://ehudreiter.com/2026/06/26/future-of-nlg-evaluation-2/", "published_at": "2026-06-26T08:22:09+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee548002acc", "title": "I am worried by NLP research culture", "url": "https://ehudreiter.com/2026/06/08/nlp-research-culture/", "published_at": "2026-06-08T08:04:47+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee5483b595c", "title": "Software engineering of prompts", "url": "https://ehudreiter.com/2026/05/20/software-engineering-of-prompts/", "published_at": "2026-05-20T08:16:42+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee548dbff66", "title": "AI and CS Teaching", "url": "https://ehudreiter.com/2026/05/05/ai-and-cs-teaching/", "published_at": "2026-05-05T06:35:22+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee549c6dee6", "title": "Achieving my vision of personal AI health assistants", "url": "https://ehudreiter.com/2026/04/20/achieving-personal-ai-health-assistants/", "published_at": "2026-04-20T15:26:24+00:00" }, { "id": "01a07df2-b62a-73e8-9118-cee54ab7b087", "title": "Real-world safety and harms from patient-facing LLMs", "url": "https://ehudreiter.com/2026/04/06/real-world-safety-in-patient-facing-llms/", "published_at": "2026-04-06T07:36:37+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b456ebbf3cc", "title": "Comparing performance of LLMs is not very interesting", "url": "https://ehudreiter.com/2026/03/24/comparing-performance-of-llms-is-not-very-interesting/", "published_at": "2026-03-24T07:50:09+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b456f6d8abc", "title": "Please follow the rules for ARR/ACL papers", "url": "https://ehudreiter.com/2026/03/16/please-follow-the-rules-for-arr-acl-papers/", "published_at": "2026-03-16T09:08:50+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b456fa8e83d", "title": "Questions from readers of my book", "url": "https://ehudreiter.com/2026/03/03/questions-from-readers-of-my-book/", "published_at": "2026-03-03T09:16:21+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b456fcc08b0", "title": "Dont ignore omissions!", "url": "https://ehudreiter.com/2026/02/11/dont-ignore-omissions/", "published_at": "2026-02-11T09:39:15+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b4570a629fc", "title": "My Eureka moments in research", "url": "https://ehudreiter.com/2026/01/30/my-eureka-moments-in-research/", "published_at": "2026-01-30T08:57:11+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b457101b876", "title": "Lets use AI to help people manage illness", "url": "https://ehudreiter.com/2026/01/19/lets-use-ai-to-help-people-manage-illness/", "published_at": "2026-01-19T09:16:22+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b4571c4500d", "title": "Retirement Plans: Travel and some academics", "url": "https://ehudreiter.com/2026/01/06/retirement-plans-travel-and-some-academics/", "published_at": "2026-01-06T08:20:54+00:00" }, { "id": "01a0c50a-3b71-72a8-9c70-6b4571f91d49", "title": "Do a sanity check on your experiments", "url": "https://ehudreiter.com/2025/12/22/do-a-sanity-check-on-your-experiments/", "published_at": "2025-12-22T08:57:48+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af001bc8c8", "title": "Do LLMs cheat on benchmarks", "url": "https://ehudreiter.com/2025/12/08/do-llms-cheat-on-benchmarks/", "published_at": "2025-12-08T06:40:45+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af00fabe72", "title": "Hard to Change Poor Research Culture", "url": "https://ehudreiter.com/2025/11/24/hard-to-change-poor-research-culture/", "published_at": "2025-11-24T09:04:48+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af01afdcb6", "title": "Understanding what users want from NLG", "url": "https://ehudreiter.com/2025/11/06/understanding-what-users-want-from-nlg/", "published_at": "2025-11-06T07:22:14+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af02179296", "title": "Most common uses of AI in Healthcare", "url": "https://ehudreiter.com/2025/10/21/most-common-uses-of-ai-in-healthcare/", "published_at": "2025-10-21T06:16:15+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af0305411c", "title": "Good diagrams for research papers", "url": "https://ehudreiter.com/2025/10/08/good-diagrams-for-research-papers/", "published_at": "2025-10-08T08:22:50+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af03ffe7b9", "title": "Reflections on blogging", "url": "https://ehudreiter.com/2025/09/23/reflections-on-blogging/", "published_at": "2025-09-23T07:50:18+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af04d45020", "title": "Defining hallucination is not straightforward", "url": "https://ehudreiter.com/2025/09/10/defining-hallucination-is-not-straightforward/", "published_at": "2025-09-10T05:17:50+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af050b294c", "title": "Encouraging safer driving with NLG apps", "url": "https://ehudreiter.com/2025/09/03/encouraging-safer-driving-with-nlg-apps/", "published_at": "2025-09-03T05:40:12+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af050f2b33", "title": "I hate pay-to-publish", "url": "https://ehudreiter.com/2025/08/19/i-hate-pay-to-publish/", "published_at": "2025-08-19T08:28:23+00:00" }, { "id": "01a0c51a-45bb-70af-a13a-56af06070708", "title": "More on evaluating impact", "url": "https://ehudreiter.com/2025/08/05/more-on-evaluating-impact/", "published_at": "2025-08-05T06:35:06+00:00" }, { "id": "01a0c525-46b8-712e-aa1c-69247a2368eb", "title": "Cycling in Netherlands", "url": "https://ehudreiter.com/2025/07/13/cycling-in-netherlands/", "published_at": "2025-07-13T10:32:58+00:00" }, { "id": "01a0c525-46b8-712e-aa1c-69247aa2f1b3", "title": "Patients want to know what information an AI model considers", "url": "https://ehudreiter.com/2025/06/25/patients-want-to-know-what-information-an-ai-model-considers/", "published_at": "2025-06-25T05:52:45+00:00" }, { "id": "01a0c525-46b8-712e-aa1c-69247b5b1860", "title": "The Aberdeen NLP Research Group", "url": "https://ehudreiter.com/2025/06/05/the-aberdeen-nlp-research-group/", "published_at": "2025-06-05T06:52:13+00:00" }, { "id": "01a0c525-46b8-712e-aa1c-69247bb86060", "title": "Key messages from my NLG book", "url": "https://ehudreiter.com/2025/05/14/key-messages-from-my-nlg-book/", "published_at": "2025-05-14T10:11:07+00:00" }, { "id": "01a0c525-46b9-73b9-b9d1-7002a37644bf", "title": "Even good leaderboards may not be useful, because they are gamed", "url": "https://ehudreiter.com/2025/05/05/good-leaderboards-gamed/", "published_at": "2025-05-05T08:18:38+00:00" }, { "id": "01a0c525-46b9-73b9-b9d1-7002a44dfffe", "title": "Examples of evaluating real-world impact", "url": "https://ehudreiter.com/2025/04/08/examples-of-evaluating-real-world-impact/", "published_at": "2025-04-08T06:59:07+00:00" }, { "id": "01a0c525-46ba-73c7-95d2-26debdf85def", "title": "Benchmarks distract us from what matters", "url": "https://ehudreiter.com/2025/03/26/benchmarks-distract-us-from-what-matters/", "published_at": "2025-03-26T08:25:29+00:00" }, { "id": "01a0c525-46ba-73c7-95d2-26debeb3f765", "title": "People do not understand how LLMs can/cannot help them", "url": "https://ehudreiter.com/2025/03/13/people-dont-understand-how-llms-can-help-them/", "published_at": "2025-03-13T07:51:03+00:00" }, { "id": "01a0c525-46ba-73c7-95d2-26debef227ac", "title": "Improving Bayesian Networks", "url": "https://ehudreiter.com/2025/03/03/improving-bayesian-networks/", "published_at": "2025-03-03T07:26:52+00:00" }, { "id": "01a0c525-46ba-73c7-95d2-26debf053384", "title": "I want a benchmark for emotional upset", "url": "https://ehudreiter.com/2025/02/17/benchmark-for-emotional-upset/", "published_at": "2025-02-17T12:52:26+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d342a6d13", "title": "NLG Evaluation 2025 vs 2015: much improved but needs to be better", "url": "https://ehudreiter.com/2025/02/04/nlg-evaluation-2025-vs-2015/", "published_at": "2025-02-04T06:24:38+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d34af95cc", "title": "Vision: AI personal health assistants", "url": "https://ehudreiter.com/2025/01/23/vision-ai-personal-health-assistants/", "published_at": "2025-01-23T07:24:40+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d34e6301a", "title": "Do LLM coding benchmarks measure real-world utility?", "url": "https://ehudreiter.com/2025/01/13/do-llm-coding-benchmarks-measure-real-world-utility/", "published_at": "2025-01-13T09:26:26+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d35e28c21", "title": "We need better LLM benchmarks", "url": "https://ehudreiter.com/2025/01/03/we-need-better-llm-benchmarks/", "published_at": "2025-01-03T09:02:41+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d36306a7d", "title": "Do LLM benchmarks ignore NLG?", "url": "https://ehudreiter.com/2024/12/26/do-llm-benchmarks-ignore-nlg/", "published_at": "2024-12-26T09:26:21+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d36b6e4b7", "title": "Interesting papers in 2024", "url": "https://ehudreiter.com/2024/12/16/interesting-papers-in-2024/", "published_at": "2024-12-16T06:51:20+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d37360a6d", "title": "MQM shows the power of a gold-standard evaluation", "url": "https://ehudreiter.com/2024/12/02/mqm-shows-the-power-of-a-gold-standard-evaluation/", "published_at": "2024-12-02T06:13:19+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d3807f082", "title": "Make your next experiment a bit better", "url": "https://ehudreiter.com/2024/11/18/make-your-next-experiment-a-bit-better/", "published_at": "2024-11-18T07:12:18+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d38ca845e", "title": "My New Book on NLG", "url": "https://ehudreiter.com/2024/11/06/my-new-book-on-nlg/", "published_at": "2024-11-06T08:03:12+00:00" }, { "id": "01a0c530-c1dd-7199-9e27-0f1d39766253", "title": "Maintaining NLG Systems", "url": "https://ehudreiter.com/2024/10/21/maintaining-nlg-systems/", "published_at": "2024-10-21T06:10:06+00:00" }, { "id": "01a0c53a-dd9d-73fe-b8a5-6e1bba7d615c", "title": "Qualitative evaluation", "url": "https://ehudreiter.com/2024/10/07/qualitative-evaluation/", "published_at": "2024-10-07T06:35:45+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42deebc53a", "title": "How AI can help reform UK NHS", "url": "https://ehudreiter.com/2024/09/23/how-ai-can-help-reform-uk-nhs/", "published_at": "2024-09-23T08:31:07+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42df195373", "title": "One-day class on NLG evaluation", "url": "https://ehudreiter.com/2024/09/09/one-day-class-on-nlg-evaluation/", "published_at": "2024-09-09T05:55:41+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42dfb984f8", "title": "The latest/trendiest tech isnt always appropriate", "url": "https://ehudreiter.com/2024/08/26/the-latest-trendiest-tech-isnt-always-appropriate/", "published_at": "2024-08-26T06:48:21+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42dfd738e6", "title": "Heading towards Retirement", "url": "https://ehudreiter.com/2024/08/09/heading-towards-retirement/", "published_at": "2024-08-09T07:16:57+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42e0536ca4", "title": "Why is adoption of AI in healthcare so slow?", "url": "https://ehudreiter.com/2024/07/23/slow-adoption-of-ai-in-healthcare/", "published_at": "2024-07-23T08:16:10+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42e09f436a", "title": "Challenges in Evaluating LLMs", "url": "https://ehudreiter.com/2024/07/10/challenges-in-evaluating-llms/", "published_at": "2024-07-10T06:37:51+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42e0f391e7", "title": "Cycling from Perth to Preston", "url": "https://ehudreiter.com/2024/07/01/cycling-from-perth-to-preston/", "published_at": "2024-07-01T06:36:58+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42e13b1bd4", "title": "Can LLM-based eval replace human evaluation?", "url": "https://ehudreiter.com/2024/06/11/llm-vs-human-eval/", "published_at": "2024-06-11T06:56:41+00:00" }, { "id": "01a0c53a-dd9e-7360-b6f9-6f42e1805a15", "title": "Well structured input data helps LLMs", "url": "https://ehudreiter.com/2024/06/03/well-structured-input-data-helps-llms/", "published_at": "2024-06-03T06:39:51+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bd51b54b", "title": "Human eval: Subjects must understand the task", "url": "https://ehudreiter.com/2024/05/28/human-eval-subjects-must-understand-the-task/", "published_at": "2024-05-28T08:12:44+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bdb19b46", "title": "We can learn from the past in AI/Medicine", "url": "https://ehudreiter.com/2024/05/06/learn-past-in-ai-medicine/", "published_at": "2024-05-06T06:48:55+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bdb6fa65", "title": "Real-world usage of LLMs in Journalism", "url": "https://ehudreiter.com/2024/04/23/usage-llms-journalism/", "published_at": "2024-04-23T06:43:28+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bdc69d1a", "title": "Ten tips on doing a good evaluation", "url": "https://ehudreiter.com/2024/04/08/ten-tips-on-doing-a-good-evaluation/", "published_at": "2024-04-08T07:03:22+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72be6dd3cd", "title": "Explaining complex information to patients", "url": "https://ehudreiter.com/2024/03/25/explaining-complex-information-to-patients/", "published_at": "2024-03-25T09:40:15+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72beb87afa", "title": "I’m very worried about data contamination", "url": "https://ehudreiter.com/2024/03/12/data-contamination-worries/", "published_at": "2024-03-12T08:36:42+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bf1f767c", "title": "In 2019 LM output was fluent but not trustworthy: still true in 2024", "url": "https://ehudreiter.com/2024/02/27/lm-output-still-fluent-not-trustworthy/", "published_at": "2024-02-27T08:25:45+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bf623f4d", "title": "Communicating uncertainty to non-experts", "url": "https://ehudreiter.com/2024/02/13/communicating-uncertainty-to-non-experts/", "published_at": "2024-02-13T07:15:18+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72bf8df976", "title": "Systematic Reviews in NLP", "url": "https://ehudreiter.com/2024/01/31/systematic-reviews-in-nlp/", "published_at": "2024-01-31T07:52:16+00:00" }, { "id": "01a0c544-0d79-710f-b294-8a72c013f03e", "title": "Common Flaws in NLP Evaluation Experiments", "url": "https://ehudreiter.com/2024/01/15/common-flaws-in-nlp-evaluation-experiments/", "published_at": "2024-01-15T08:06:34+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b6437806", "title": "Can LLMs make medicine safer?", "url": "https://ehudreiter.com/2023/12/28/can-llms-make-medicine-safer/", "published_at": "2023-12-28T13:16:35+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b6b035f8", "title": "NLP has become much more interesting!", "url": "https://ehudreiter.com/2023/12/20/nlp-has-become-much-more-interesting/", "published_at": "2023-12-20T07:28:03+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b795f7c4", "title": "What LLMs cannot do", "url": "https://ehudreiter.com/2023/12/11/what-llms-cannot-do/", "published_at": "2023-12-11T08:13:36+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b8655d85", "title": "Indifference to climate change makes me angry", "url": "https://ehudreiter.com/2023/11/27/climate-change-makes-me-angry/", "published_at": "2023-11-27T07:58:12+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b8fb6b16", "title": "We should evaluate real-world impact!", "url": "https://ehudreiter.com/2023/11/13/evaluate-real-world-impact/", "published_at": "2023-11-13T07:42:05+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1b9692e37", "title": "A bad way to measure hallucination", "url": "https://ehudreiter.com/2023/10/31/a-bad-way-to-measure-hallucination/", "published_at": "2023-10-31T08:18:27+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1ba19b91c", "title": "ChatGPT error or human error?", "url": "https://ehudreiter.com/2023/10/19/chatgpt-error-or-human-error/", "published_at": "2023-10-19T06:29:51+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1baf109e7", "title": "Peer Review Has Improved My Papers", "url": "https://ehudreiter.com/2023/10/10/peer-review-has-improved-my-papers/", "published_at": "2023-10-10T07:08:50+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1bb8842a7", "title": "NLG texts should not upset people", "url": "https://ehudreiter.com/2023/09/26/nlg-texts-should-not-upset-people/", "published_at": "2023-09-26T06:45:16+00:00" }, { "id": "01a0c54c-e222-73f3-a84f-05b1bc1a48f9", "title": "There are many types of human evaluation!", "url": "https://ehudreiter.com/2023/09/13/many-types-of-human-evaluation/", "published_at": "2023-09-13T06:02:09+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0b5cf104", "title": "Problems in using LLMs in commercial products", "url": "https://ehudreiter.com/2023/08/24/llm-problems-in-commercial-products/", "published_at": "2023-08-24T08:41:36+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0c3bfd23", "title": "My MSc students evaluate chatGPT", "url": "https://ehudreiter.com/2023/08/16/my-msc-students-evaluate-chatgpt/", "published_at": "2023-08-16T06:39:22+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0cf1c189", "title": "LLM hype brings memories of IBM Watson", "url": "https://ehudreiter.com/2023/07/31/llm-hype-brings-memories-of-ibm-watson/", "published_at": "2023-07-31T07:36:38+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0db7a001", "title": "Are Experts Needed in Human Evaluation?", "url": "https://ehudreiter.com/2023/07/10/are-experts-needed-in-human-evaluation/", "published_at": "2023-07-10T06:27:47+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0e9f0f2e", "title": "LLMs and Data-to-text", "url": "https://ehudreiter.com/2023/06/29/llms-and-data-to-text/", "published_at": "2023-06-29T06:37:42+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0f752379", "title": "Evaluation: Plan ahead, details matter, keep it simple, pilot, be careful", "url": "https://ehudreiter.com/2023/06/21/evaluation-plan-ahead-details-matter-keep-it-simple-pilot-be-careful/", "published_at": "2023-06-21T08:24:04+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0f9aaad1", "title": "ACL vs TACL Reviewing", "url": "https://ehudreiter.com/2023/06/06/acl-vs-tacl-reviewing/", "published_at": "2023-06-06T07:10:37+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c0fd8927a", "title": "Future of NLG evaluation: LLMs and high quality human eval?", "url": "https://ehudreiter.com/2023/05/22/future-of-nlg-evaluation/", "published_at": "2023-05-22T06:08:23+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c1006bc4f", "title": "Limits of pre-publication reviewing", "url": "https://ehudreiter.com/2023/05/09/limits-of-reviewing-papers/", "published_at": "2023-05-09T06:03:26+00:00" }, { "id": "01a0c555-5515-7352-9546-6f7c10bcc5a4", "title": "Unresponsive Authors and Experimental Flaws", "url": "https://ehudreiter.com/2023/05/03/unresponsive-authors-and-experimental-flaws/", "published_at": "2023-05-03T06:03:23+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ea5e8cb3", "title": "chatGPT in Health: Exciting if we ignore the hype", "url": "https://ehudreiter.com/2023/04/09/chatgpt-in-health-exciting-if-we-ignore-the-hype/", "published_at": "2023-04-09T17:11:23+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996eb56401f", "title": "Evaluating chatGPT", "url": "https://ehudreiter.com/2023/04/04/evaluating-chatgpt/", "published_at": "2023-04-04T08:22:42+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ec39bf39", "title": "Does chatGPT make leaderboards less meaningful?", "url": "https://ehudreiter.com/2023/03/27/chatgpt-and-leaderboard/", "published_at": "2023-03-27T06:21:07+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ec50f586", "title": "Could some NLP research be fraudulent?", "url": "https://ehudreiter.com/2023/03/07/fraud-in-nlp-research/", "published_at": "2023-03-07T07:47:43+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ed406845", "title": "What Should Academic NLP Researchers Focus on?", "url": "https://ehudreiter.com/2023/02/28/academic-nlp-researchers-focus/", "published_at": "2023-02-28T07:48:54+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996edc8fb6d", "title": "How accurate do chatGPT texts need to be?", "url": "https://ehudreiter.com/2023/02/14/required-accuracy-chatgpt/", "published_at": "2023-02-14T07:47:45+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ee6e78f7", "title": "Evaluating factual accuracy in complex data-to-text", "url": "https://ehudreiter.com/2023/02/07/evaluating-factual-accuracy-in-complex-data-to-text/", "published_at": "2023-02-07T07:32:26+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ee9dc806", "title": "Can ChatGPT do Data-to-Text?", "url": "https://ehudreiter.com/2023/01/23/chatgpt-data-to-text/", "published_at": "2023-01-23T07:18:20+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ef2fd342", "title": "Texts can be accurate but still not appropriate", "url": "https://ehudreiter.com/2023/01/16/texts-accurate-but-not-appropriate/", "published_at": "2023-01-16T07:03:41+00:00" }, { "id": "01a0c55d-a1b2-7110-9951-6996ef80fc66", "title": "chatGPT: Great science, unclear commercials, hate the hype", "url": "https://ehudreiter.com/2022/12/29/chatgpt-great-science-unclear-commercials-hate-the-hype/", "published_at": "2022-12-29T08:42:15+00:00" } ] posts Claim your blog
Back to Ehud Reiter
Blog · corpus.blog/blogs/ehudreiter.com/posts

Ehud Reiter

ehudreiter.com

2026

Localising NLG content in a driving feedback systemoriginal ↗
17 Sept 2026
Desk rejection is an unfortunate necessityoriginal ↗
8 Sept 2026
Cycling in Northern Irelandoriginal ↗
25 Aug 2026
AI for Healthcare: Great benchmark scores but minimal impactoriginal ↗
5 Aug 2026
Memories of being an NL researcher in 1990original ↗
20 Jul 2026
What is the purpose of ACL conferences?original ↗
9 Jul 2026
Future of NLG evaluationoriginal ↗
26 Jun 2026
I am worried by NLP research cultureoriginal ↗
8 Jun 2026
Software engineering of promptsoriginal ↗
20 May 2026
AI and CS Teachingoriginal ↗
5 May 2026
Achieving my vision of personal AI health assistantsoriginal ↗
20 Apr 2026
Real-world safety and harms from patient-facing LLMsoriginal ↗
6 Apr 2026
Comparing performance of LLMs is not very interestingoriginal ↗
24 Mar 2026
Please follow the rules for ARR/ACL papersoriginal ↗
16 Mar 2026
Questions from readers of my bookoriginal ↗
3 Mar 2026
Dont ignore omissions!original ↗
11 Feb 2026
My Eureka moments in researchoriginal ↗
30 Jan 2026
Lets use AI to help people manage illnessoriginal ↗
19 Jan 2026
Retirement Plans: Travel and some academicsoriginal ↗
6 Jan 2026

2025

Do a sanity check on your experimentsoriginal ↗
22 Dec 2025
Do LLMs cheat on benchmarksoriginal ↗
8 Dec 2025
Hard to Change Poor Research Cultureoriginal ↗
24 Nov 2025
Understanding what users want from NLGoriginal ↗
6 Nov 2025
Most common uses of AI in Healthcareoriginal ↗
21 Oct 2025
Good diagrams for research papersoriginal ↗
8 Oct 2025
Reflections on bloggingoriginal ↗
23 Sept 2025
Defining hallucination is not straightforwardoriginal ↗
10 Sept 2025
Encouraging safer driving with NLG appsoriginal ↗
3 Sept 2025
I hate pay-to-publishoriginal ↗
19 Aug 2025
More on evaluating impactoriginal ↗
5 Aug 2025
Cycling in Netherlandsoriginal ↗
13 Jul 2025
Patients want to know what information an AI model considersoriginal ↗
25 Jun 2025
The Aberdeen NLP Research Grouporiginal ↗
5 Jun 2025
Key messages from my NLG bookoriginal ↗
14 May 2025
Even good leaderboards may not be useful, because they are gamedoriginal ↗
5 May 2025
Examples of evaluating real-world impactoriginal ↗
8 Apr 2025
Benchmarks distract us from what mattersoriginal ↗
26 Mar 2025
People do not understand how LLMs can/cannot help themoriginal ↗
13 Mar 2025
Improving Bayesian Networksoriginal ↗
3 Mar 2025
I want a benchmark for emotional upsetoriginal ↗
17 Feb 2025
NLG Evaluation 2025 vs 2015: much improved but needs to be betteroriginal ↗
4 Feb 2025
Vision: AI personal health assistantsoriginal ↗
23 Jan 2025
Do LLM coding benchmarks measure real-world utility?original ↗
13 Jan 2025
We need better LLM benchmarksoriginal ↗
3 Jan 2025

2024

Do LLM benchmarks ignore NLG?original ↗
26 Dec 2024
Interesting papers in 2024original ↗
16 Dec 2024
MQM shows the power of a gold-standard evaluationoriginal ↗
2 Dec 2024
Make your next experiment a bit betteroriginal ↗
18 Nov 2024
My New Book on NLGoriginal ↗
6 Nov 2024
Maintaining NLG Systemsoriginal ↗
21 Oct 2024
Qualitative evaluationoriginal ↗
7 Oct 2024
How AI can help reform UK NHSoriginal ↗
23 Sept 2024
One-day class on NLG evaluationoriginal ↗
9 Sept 2024
The latest/trendiest tech isnt always appropriateoriginal ↗
26 Aug 2024
Heading towards Retirementoriginal ↗
9 Aug 2024
Why is adoption of AI in healthcare so slow?original ↗
23 Jul 2024
Challenges in Evaluating LLMsoriginal ↗
10 Jul 2024
Cycling from Perth to Prestonoriginal ↗
1 Jul 2024
Can LLM-based eval replace human evaluation?original ↗
11 Jun 2024
Well structured input data helps LLMsoriginal ↗
3 Jun 2024
Human eval: Subjects must understand the taskoriginal ↗
28 May 2024
We can learn from the past in AI/Medicineoriginal ↗
6 May 2024
Real-world usage of LLMs in Journalismoriginal ↗
23 Apr 2024
Ten tips on doing a good evaluationoriginal ↗
8 Apr 2024
Explaining complex information to patientsoriginal ↗
25 Mar 2024
I’m very worried about data contaminationoriginal ↗
12 Mar 2024
In 2019 LM output was fluent but not trustworthy: still true in 2024original ↗
27 Feb 2024
Communicating uncertainty to non-expertsoriginal ↗
13 Feb 2024
Systematic Reviews in NLPoriginal ↗
31 Jan 2024
Common Flaws in NLP Evaluation Experimentsoriginal ↗
15 Jan 2024

2023

Can LLMs make medicine safer?original ↗
28 Dec 2023
NLP has become much more interesting!original ↗
20 Dec 2023
What LLMs cannot dooriginal ↗
11 Dec 2023
Indifference to climate change makes me angryoriginal ↗
27 Nov 2023
We should evaluate real-world impact!original ↗
13 Nov 2023
A bad way to measure hallucinationoriginal ↗
31 Oct 2023
ChatGPT error or human error?original ↗
19 Oct 2023
Peer Review Has Improved My Papersoriginal ↗
10 Oct 2023
NLG texts should not upset peopleoriginal ↗
26 Sept 2023
There are many types of human evaluation!original ↗
13 Sept 2023
Problems in using LLMs in commercial productsoriginal ↗
24 Aug 2023
My MSc students evaluate chatGPToriginal ↗
16 Aug 2023
LLM hype brings memories of IBM Watsonoriginal ↗
31 Jul 2023
Are Experts Needed in Human Evaluation?original ↗
10 Jul 2023
LLMs and Data-to-textoriginal ↗
29 Jun 2023
Evaluation: Plan ahead, details matter, keep it simple, pilot, be carefuloriginal ↗
21 Jun 2023
ACL vs TACL Reviewingoriginal ↗
6 Jun 2023
Future of NLG evaluation: LLMs and high quality human eval?original ↗
22 May 2023
Limits of pre-publication reviewingoriginal ↗
9 May 2023
Unresponsive Authors and Experimental Flawsoriginal ↗
3 May 2023
chatGPT in Health: Exciting if we ignore the hypeoriginal ↗
9 Apr 2023
Evaluating chatGPToriginal ↗
4 Apr 2023
Does chatGPT make leaderboards less meaningful?original ↗
27 Mar 2023
Could some NLP research be fraudulent?original ↗
7 Mar 2023
What Should Academic NLP Researchers Focus on?original ↗
28 Feb 2023
How accurate do chatGPT texts need to be?original ↗
14 Feb 2023
Evaluating factual accuracy in complex data-to-textoriginal ↗
7 Feb 2023
Can ChatGPT do Data-to-Text?original ↗
23 Jan 2023
Texts can be accurate but still not appropriateoriginal ↗
16 Jan 2023

2022

chatGPT: Great science, unclear commercials, hate the hypeoriginal ↗
29 Dec 2022
Next