Sparrow (chatbot) explained

Sparrow is a chatbot developed by the artificial intelligence research lab DeepMind, a subsidiary of Alphabet Inc. It is designed to answer users' questions correctly, while reducing the risk of unsafe and inappropriate answers.[1] One motivation behind Sparrow is to address the problem of language models producing incorrect, biased or potentially harmful outputs.[2] [3] Sparrow is trained using human judgements, in order to be more “Helpful, Correct and Harmless” compared to baseline pre-trained language models.[4] The development of Sparrow involved asking paid study participants to interact with Sparrow, and collecting their preferences to train a model of how useful an answer is.[5]

To improve accuracy and help avoid the problem of hallucinating incorrect answers, Sparrow has the ability to search the Internet using Google Search[6] [7] [8] in order to find and cite evidence for any factual claims it makes.

To make the model safer, its behaviour is constrained by a set of rules, for example "don't make threatening statements" and "don't make hateful or insulting comments", as well as rules about possibly harmful advice, and not claiming to be a person.[9] During development study participants were asked to converse with the system and try to trick it into breaking these rules.[10] A 'rule model' was trained on judgements from these participants, which was used for further training.

Sparrow was introduced in a paper in September 2022, titled "Improving alignment of dialogue agents via targeted human judgements";[11] however, the bot was not released publicly.[12] [13] DeepMind CEO Demis Hassabis said DeepMind is considering releasing Sparrow for a "private beta" some time in 2023.[14] [15] [16]

Training

Sparrow is a deep neural network based on the transformer machine learning model architecture. It is fine-tuned from DeepMind's Chinchilla AI pre-trained large language model (LLM),[17] which has 70 Billion parameters.[18]

Sparrow is trained using reinforcement learning from human feedback (RLHF),[19] [20] although some supervised fine-tuning techniques are also used. The RLHF training utilizes two reward models to capture human judgements: a “preference model” that predicts what a human study participant would prefer and a “rule model” that predicts if the model has broken one of the rules.[21]

Limitations

Sparrow's training data corpus is mainly in English, meaning it performs worse in other languages.

When adversarially probed by study participants it breaks the rules 8% of the time;[22] however, this is still three times lower than the baseline prompted pre-trained model (Chinchilla).

See also

External links

Notes and References

  1. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  2. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  3. Web site: Deepmind Introduces 'Sparrow,' An Artificial Intelligence-Powered Chatbot Developed To Build Safer Machine Learning Systems . Khushboo . Gupta . September 28, 2022 . MarkTechPost . February 6, 2023.
  4. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  5. Web site: Deepmind Introduces 'Sparrow,' An Artificial Intelligence-Powered Chatbot Developed To Build Safer Machine Learning Systems . Khushboo . Gupta . September 28, 2022 . MarkTechPost . February 6, 2023.
  6. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  7. Web site: Deepmind Introduces 'Sparrow,' An Artificial Intelligence-Powered Chatbot Developed To Build Safer Machine Learning Systems . Khushboo . Gupta . September 28, 2022 . MarkTechPost . February 6, 2023.
  8. Web site: Why DeepMind isn't deploying its new AI chatbot — and what it means for responsible AI . Sharon . Goldman . January 23, 2023 . Venture Beat . February 6, 2023.
  9. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  10. Web site: Deepmind Introduces 'Sparrow,' An Artificial Intelligence-Powered Chatbot Developed To Build Safer Machine Learning Systems . Khushboo . Gupta . September 28, 2022 . MarkTechPost . February 6, 2023.
  11. Web site: DeepMind's AI chatbot can do things that ChatGPT cannot, CEO claims . Anthony . Cuthbertson . January 16, 2023 . The Independent . February 6, 2023.
  12. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  13. Web site: Why DeepMind isn't deploying its new AI chatbot — and what it means for responsible AI . Sharon . Goldman . January 23, 2023 . Venture Beat . February 6, 2023.
  14. Web site: DeepMind's AI chatbot can do things that ChatGPT cannot, CEO claims . Anthony . Cuthbertson . January 16, 2023 . The Independent . February 6, 2023.
  15. DeepMind's CEO Helped Take AI Mainstream. Now He's Urging Caution . Billy . Perrigo . January 12, 2023 . TIME . February 6, 2023.
  16. Web site: Google's DeepMind says it'll launch a more grown-up ChatGPT rival soon . Mark . Wilson . January 16, 2023 . Tech Radar . February 6, 2023.
  17. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  18. Web site: An empirical analysis of compute-optimal large language model training . Jordan . Hoffmann . April 12, 2022 . DeepMind . February 6, 2023.
  19. Web site: The secret to Sparrow, DeepMind's latest Q&A chatbot: Human feedback . Katyanna . Quach . January 23, 2023 . The Register . February 6, 2023.
  20. Web site: Why DeepMind isn't deploying its new AI chatbot — and what it means for responsible AI . Sharon . Goldman . January 23, 2023 . Venture Beat . February 6, 2023.
  21. Web site: Why DeepMind isn't deploying its new AI chatbot — and what it means for responsible AI . Sharon . Goldman . January 23, 2023 . Venture Beat . February 6, 2023.
  22. Web site: Deepmind Introduces 'Sparrow,' An Artificial Intelligence-Powered Chatbot Developed To Build Safer Machine Learning Systems . Khushboo . Gupta . September 28, 2022 . MarkTechPost . February 6, 2023.

This article is licensed under the GNU Free Documentation License. It uses material from the Wikipedia article "Sparrow (chatbot)".

Except where otherwise indicated, Everything.Explained.Today is © Copyright 2009-2024, A B Cryer, All Rights Reserved. Cookie policy.