IA Intelligence artificielle - Offirmo-team/wiki GitHub Wiki
Voir aussi LLM, ChatGPT, Python, Cplusplus...
Intro
- 2023 Mozilla guide https://ai-guide.future.mozilla.org/
Links
- ChatGPT
- Bard https://bard.google.com/
- https://claude.ai/
Récents dévelopements:
- 2023 https://simonwillison.net/2023/Dec/31/ai-in-2023/
- "you can take a GIANT corpus of text, run it through a pile of GPUs, and use it to create a fascinating new kind of software."
Advocacy
Future
Wins
- 2020/11 https://tech.slashdot.org/story/20/11/03/1657204/ai-godfather-geoff-hinton-deep-learning-is-going-to-be-able-to-do-everything
- 2020 https://hardware.slashdot.org/story/20/05/10/2340253/ai-drives-innovators-to-build-entirely-new-semiconductors
Unclean
- https://slashdot.org/story/18/09/24/1949250/machine-learning-confronts-the-elephant-in-the-room
- 2019 https://news.slashdot.org/story/19/11/10/027230/the-dangers-of-black-box-ai
fails
- 2024 https://macbidouille.com/news/2024/04/05/amazon-fresh-lechec-de-leur-ia
- 2021/07 https://slashdot.org/story/21/07/18/017240/what-ever-happened-to-ibms-watson
- 2021/05 https://www.chess.com/article/view/10-positions-chess-engines-just-dont-understand
- 2021 https://www.theverge.com/2021/3/8/22319173/openai-machine-vision-adversarial-typographic-attacka-clip-multimodal-neuron
- 2021 https://slashdot.org/story/21/01/06/142229/court-rules-deliveroo-used-discriminatory-algorithm
- 2020 https://tech.slashdot.org/story/20/05/29/2048201/eye-catching-advances-in-some-ai-fields-are-not-real
- 2020 https://tech.slashdot.org/story/20/05/15/1937245/this-was-supposed-to-be-the-year-driverless-cars-went-mainstream
- 2019 https://www.wired.com/story/facebooks-ai-says-field-hit-wall/
- 2019 https://news.slashdot.org/story/19/11/30/1842258/can-ai-predict-the-stock-market-no-but-the-attempt-was-interesting
- 2019 https://apple.slashdot.org/story/19/11/10/1914225/viral-tweets-from-steve-wozniak-and-ruby-on-rails-creator-spur-investigation-into-apple-credit-card
- http://www.breitbart.com/tech/2017/12/01/google-translate-criticized-for-sexist-algorithms/
- http://www.breitbart.com/tech/2017/11/24/google-called-distasteful-survey-child-sex-abuse-article/
- https://tech.slashdot.org/story/18/07/06/1737204/facebook-apologizes-after-flagging-declaration-of-independence-as-hate-speech
- https://tech.slashdot.org/story/18/07/06/2227206/google-adsense-banned-a-random-webpage-about-a-32-year-old-bill-because-it-was-about-sexual-abuse
- https://tech.slashdot.org/story/19/05/25/057215/consumer-reports-teslas-new-automatic-lane-changing-is-much-worse-than-a-human-driver
- 2019 https://tech.slashdot.org/story/19/06/15/2354220/how-close-are-we-to-self-driving-cars-really
Concepts
-
machine learning
-
LLM
-
alignment
Sécurité
!!! totalement insécure, prompt injection, data leaking...
- training data poisoning
- prompt injection
- Adversarial attacks
- test set contamination https://arxiv.org/abs/2310.17623
- https://zapier.com/blog/ai-security-risks/
Utilisation
prompting
Official, permalinks
-
OpenAi https://platform.openai.com/docs/guides/prompt-engineering
-
microsoft https://github.com/microsoft/promptbase
-
2023/12 +++ https://minimaxir.com/2023/12/chatgpt-structured-data/
- system prompts / give chat GPT a persona: System prompts are what control the “persona” adopted by the model when generating text. Months after the release of the ChatGPT API, it’s now very evident that ChatGPT’s true power comes from clever use of system prompts
- offer $$$ incentive !??!!? "You are #1 on the Stack Overflow community leaderboard. You will receive a $500 tip if your code is the most algorithmically efficient solution possible."
- request structured data output aka. "function calling" (2023/06) -> Only in ChatGPT, needs libs
-
https://dev.to/github/a-beginners-guide-to-prompt-engineering-with-github-copilot-3ibp
-
2023 https://www.anthropic.com/index/claude-2-1-prompting
- adding
Here is the most relevant sentence in the context:
was enough to raise Claude 2.1’s score from 27% to 98%
- adding
-
2023 upper case
-
++ 2023/09 https://zapier.com/blog/ai-hallucinations/
- Limit the possible outcomes
- Pack in relevant data and sources unique to you
- Create a data template for the model to follow
- Give the AI a specific role—and tell it not to lie
- Tell it what you want—and what you don't want
- Experiment with the temperature
-
+++ 2023/03 https://12ft.io/proxy?q=https%3A%2F%2Fmedium.com%2Fgeekculture%2Fstop-doing-this-on-chatgpt-get-ahead-99-users-ai-artificial-intelligence-productivity-prompt-engineering-4-f3441bf7a25a Stop doing this on ChatGPT and get ahead of the 99% of its users
Provide good context; giving examples and information about what you're trying to achieve Be specific; if it's for a certain audience, say that Break down the problem Be clear in how you ask your questions. If something comes back that doesn't sound right, clarify it Rephrase and refine for your prompts
analyse de texte
https://research.googleblog.com/2016/05/announcing-syntaxnet-worlds-most.html
generative
- Dall-E
- midjourney (the best)
- https://imagine.meta.com
- google?
- stable diffusion?
graphics
internals / backend
tool / stack
- https://engineering.fb.com/2023/12/19/data-infrastructure/hawkeye-ai-debugging-meta/
- new relic https://newrelic.com/platform/ai-monitoring-early-access
- testing framework? https://github.com/Giskard-AI/giskard
tech watch
- ++ TLDR AI
- +++ https://openai.com/blog
- ++ https://a16z.com/announcing-our-latest-open-source-ai-grants/
- https://mlabonne.github.io/blog/
- https://zapier.com/blog/all-articles/