Veille technique à relayer #59
Description
ChatGPT & IA
- https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/
- usage educatif: https://www-technologyreview-com.cdn.ampproject.org/c/s/www.technologyreview.com/2023/04/06/1071059/chatgpt-change-not-destroy-education-openai/amp/
- https://www.bbc.com/future/article/20230405-why-ai-is-becoming-impossible-for-humans-to-understand?ocid=global_future_rss
- https://www.databricks.com/blog/2023/04/12/dolly-first-open-commercially-viable-instruction-tuned-llm
- cnil italienne autoriese à nouveau chatGPT
- cnil francaise: https://www.liberation.fr/economie/economie-numerique/informations-erronees-sur-les-personnes-le-depute-eric-bothorel-saisit-la-cnil-contre-chatgpt-20230412_YHV45ISLNRD25DTULQZIV5BJ6I/
- nouvelle méthode gestion données chatGPT https://openai.com/blog/new-ways-to-manage-your-data-in-chatgpt
- chatGPT par peren https://www.peren.gouv.fr/rapports/2023-04-06_Eclairage%20sur_CHATGPT_FR.pdf
- départ ponte IA de google pour s'exprimer librement sur les dérives https://www.nytimes.com/2023/05/01/technology/ai-google-chatbot-engineer-quits-hinton.html?smid=nytcore-ios-share&referringSource=articleShare
- openassistant: https://twitter.com/Dorialexander/status/1647375634082013184?t=EXY39He7rnVUwPYt5gW8XA&s=19
- fuite données chatGPT: https://www.rfi.fr/fr/technologies/20230409-des-donn%C3%A9es-sensibles-de-samsung-divulgu%C3%A9s-sur-chatgpt-par-des-employ%C3%A9s
- the Age of Giant AI Models Is Already Over https://www.wired.com/story/openai-ceo-sam-altman-the-age-of-giant-ai-models-is-already-over/
- Rapid progress in AI is arousing fear as well as excitement. How worried should you be? https://www.linkedin.com/posts/the-economist_rapid-progress-in-ai-is-arousing-fear-as-activity-7054770700447797248-u2ib/?utm_source=share&utm_medium=member_android
- https://www.lemonde.fr/economie/article/2023/04/25/de-chatgpt-a-midjourney-les-intelligences-artificielles-generatives-s-installent-dans-les-entreprises_6170873_3234.html
- Résumé: https://twitter.com/aakashg0/status/1653050252025552898?t=0EEsiQr2VoqmwbywtRJbIA&s=19
- open llama: https://twitter.com/haoliuhl/status/1653460935200346113?t=HNKW2yAEEt2FWWZErTPN4w&s=19
IA:
-
SAM, the groundbreaking segmentation model from @Meta
is now in available in 🤗 Transformers! https://twitter.com/huggingface/status/1649102507837210641?t=2JvWRCcYgC3IvgacSE5gYw&s=19 -
midjourney x wes anderson x star wars: https://twitter.com/Benazdia/status/1652638227755302917?t=jY0bH_whfW0gDSb-Med21A&s=19
-
midjourney risque écrire histoire https://www.lemonde.fr/les-decodeurs/article/2023/05/11/les-images-creees-par-ia-et-le-risque-de-reecrire-l-histoire_6172920_4355770.html
LLM:
sécurité et privacy :
- cybersecu: https://info.deeplearning.ai/robot-metal-workers-better-pay-for-data-wranglers-building-a-south-african-ai-hub-collaborative-language-model-2
- https://openai.com/blog/our-approach-to-ai-safety
XAI:
polars
- https://pythonspeed.com/articles/polars-exploratory-data-analysis-vs-production/
- https://rpolars.github.io/articles/rpolars.html
dataviz
- site web eurostat & gridviz: https://twitter.com/mirbole01/status/1643271152213581824?t=M_c3ufdCEqXRzr20LnF36Q&s=19
- https://twitter.com/lapply/status/1646082388198293504?t=yewtMEMArOz_1sn0WqumIg&s=19
midjourney and co
Conseil lecture
-
https://twitter.com/tcarpenter216/status/1640742485348720644?t=mRfmxtse6gs4N-XqGcPdsw&s=19
-
mort moore: https://twitter.com/intel/status/1639418740243861504?s=20
-
guide rgpd dvpeurs: https://lincnil.github.io/Guide-RGPD-du-developpeur/
API
Non classé
- https://deel-ai.github.io/xplique/
- https://anglesofattack.io/Securing_AIML_Systems_in_IW_Cox.pdf
- https://www.dbpilot.io/
- https://pypi.org/project/datafusion/
- https://observablehq.com/@observablehq/data-wrangling-translations
- https://twitter.com/theshawwn/status/1638925249709240322?t=1J-Q8VuPlAd5WC0g_Mpq3w&s=19
- Google dans la course aux LLM: https://twitter.com/O42nl/status/1642204593206509568?s=20 --> fake news ?
Newsletter 13 #64
- MapPromptMonday https://github.com/MapPromptMonday/MapPromptMonday et https://twitter.com/hashtag/MapPromptMonday
- https://arxiv.org/abs/2301.04655
- https://github.blog/2023-03-22-github-copilot-x-the-ai-powered-developer-experience/
- https://twitter.com/dustinvtran/status/1639159641635717120?t=tUqtzOHFvolL7MJef-M2Cw&s=19 accroche vitesse folle
- https://www.lemonde.fr/economie/article/2023/03/29/elon-musk-et-des-centaines-d-experts-reclament-une-pause-dans-le-developpement-de-l-ia_6167461_3234.html#xtor=AL-32280270-%5Bdefault%5D-%5Bios%5D
- https://twitter.com/AndrewYNg/status/1641121451611947009?t=VyRYGWCjBeOxFOr0-bIJcQ&s=19
- https://dair-community.social/@timnitGebru/110110514253524755
- https://twitter.com/OpenAI/status/1639297361729191936?t=j1gdTehqAT8OraZRC34iqg&s=19
- CNIL italienne: https://twitter.com/gchampeau/status/1641762455972978694?t=5yKuXo1rAznOuNTxMObfNA&s=19
- https://laion.ai/blog/open-flamingo/ FLAMINGO
- https://twitter.com/lmsysorg/status/1641529839038640128?t=p4Fbp7TnrO_5IxvzNEp4uw&s=19
- https://blog.mozilla.org/en/mozilla/introducing-mozilla-ai-investing-in-trustworthy-ai/
- https://twitter.com/mattdeitke/status/1638608472525897728 qui peut peut-être être relayé en mode : ChatGPT, DALL-E... vous ont impressionné ? Les MLLM aussi ? Ce n'est pas fini, la prochaine étape sera de générer des formes en 3D
- dolly: https://www.databricks.com/blog/2023/03/24/hello-dolly-democratizing-magic-chatgpt-open-models.html
- https://www.lemonde.fr/les-decodeurs/article/2023/03/30/comment-reconnaitre-une-image-generee-par-midjourney_6167525_4355770.html#xtor=AL-32280270-%5Bdefault%5D-%5Bios%5D
- dall-E dans bing
- un article
- dockerhub change de modèle (et rétropédale)
- https://twitter.com/Docker/status/1640845221964525573?t=XWk0hanQQh_y5bzqkzrSIA&s=19
Newsletter 12 (#62)
ChatGPT
- https://github.com/microsoft/visual-chatgpt
- https://www.linkedin.com/posts/thomas-wolf-a056857_ai-opensource-chatgpt-activity-7040347000814383104-eCXz?utm_source=share&utm_medium=member_android
- https://www.wired.com/story/the-generative-ai-search-race-has-a-dirty-secret/
- https://freakonometrics.hypotheses.org/66144
- https://arstechnica.com/cars/2023/03/gm-plans-to-let-you-talk-to-your-car-with-chatgpt-knight-rider-style/
- GPT 4
- https://www.together.xyz/blog/openchatkit
Crise boursière
- https://www.forbes.fr/technologie/la-fermeture-de-svb-plus-grosse-faillite-bancaire-aux-etats-unis-depuis-2008-va-t-elle-entrainer-un-seisme-dans-le-monde-des-techs/
- https://observablehq.com/@mbostock/bank-failures
- https://home.treasury.gov/news/press-releases/jy1337
Data science
- https://mattturck.com/mad2023/
- data scientist pas de le top 20% des boulots
- long live bigdata (newsletter elixir)
- WebR: https://r-grrr.slack.com/archives/C9NDGRNSZ/p1678366364390189