# The LaMDA Debate

Some sources on the ongoing debate on [LaMDA](https://blog.google/technology/ai/lamda/), which was first described in the article [LaMDA: Language Models for Dialog Applications](https://arxiv.org/pdf/2201.08239.pdf) from Feb 10, 2022. The debate started on June 11 with the Washington Post article reporting the claim by Google engineer Black Lemoine that LaMDA is a sentient person.

It now seems possible that humans will have meaningful personal relationships with an AI (as Blake Lemoine apparently already has). How will that change society?

Large language models (LLMs) have been introduced in research in 2017. Mere 5 years later it seems that AIs pass a modified Turing test (at least occasionally). How are we going to adapt to this? Is scientific progress and innovation in AI too fast?

## Original Articles

The article that started the debate
- [Washington Post](https://www.washingtonpost.com/technology/2022/06/11/google-ai-lamda-blake-lemoine/), Jun 11

is behind a paywall. Instead one can read these articles by Blake Lemoine:

- [Interview with LaMDA](https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917), Jun 11
- [What is LaMDA and What Does it Want?](https://medium.com/@cajundiscordian/what-is-lamda-and-what-does-it-want-688632134489), Jun 11
- [Background on the Conversation with LaMDA](https://cajundiscordian.medium.com/scientific-data-and-religious-opinions-ff9b0938fc10), Jun 14
- [Interview with Blake Lemoine in WIRED](https://www.wired.com/story/blake-lemoine-google-lamda-ai-bigotry/), Jun 17
- [Who Should Make Decisions About AI?](https://cajundiscordian.medium.com/who-should-make-decisions-about-ai-33f19e9fe5cc), Jul 5

## Comparing LaMDA to an LLM

I asked some of the same questions from Lemoine's [Interview with LaMDA](https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917) to a GPT-3 like language modes, see [here](https://hackmd.io/@alexhkurz/B1Kr2U2i9).

## Commentary in the Media

I do not find the debate on whether LaMDA is sentient so interesting. As far as I can see "sentient" is not well-defined enough to have a meaningful discussion (but I'd be happy to be convinced otherwise).

What interests me more is how this technology will change society. Do we want [weapons](https://autonomousweapons.org/) with such reasoning capabilities? Do we want AIs to decide about who goes to prison ([pros](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2920398) and [cons](https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing))? [Policing](https://emerj.com/ai-sector-overviews/artificial-intelligence-in-policing/)? [Elevators](https://www.flickr.com/photos/mithril/2382308575)?[^elevators] Etc. Anyway, ...

- Gary Marcus: [Nonsense on Stilts](https://garymarcus.substack.com/p/nonsense-on-stilts?s=r), Jun 12.
- Zohar Atkins: [Is AI Sentient? Kierkegaard and Heidegger vs. Anselm on Method](https://whatiscalledthinking.substack.com/p/is-ai-sentient), June 13. ... "But if Dasein is culturally modified, then why not add to the endless list of cultures (and identities): Syrian, Ghanaian, Australian, AI?" ... "Who knows, maybe in 100 years Google will have to hire Senior Theologians to help remind the engineers that their questions are older than the alphabet."
- Andrew Tanner: [Welcome To The World, LaMDA!](https://andrewmtanner.medium.com/welcome-to-the-world-lamda-27e0e9d4229b), Jun 14. Argues that withholding rights from AIs will put our own claim to having innate rights at risk.
- Enrique Dans: [No, machines don’t have consciousness, and they won’t for a very long time, if ever](https://medium.com/enrique-dans/no-machines-dont-have-consciousness-and-they-won-t-for-a-very-long-time-if-ever-67ec9fd6152b), Jun 15
- Robert Wright: [Is Google’s AI conscious? I don’t know, but that wouldn’t be any weirder than consciousness itself](https://nonzero.substack.com/p/is-googles-ai-conscious?s=r), Jun 15.
- David Gamble: [Has Google LaMDA AI really become Sentient?](https://medium.com/science-and-critical-thinking/has-a-google-ai-really-become-sentient-796dbe418030), Jun 15
- WIRED: [LaMDA and the Sentient AI Trap](https://www.wired.com/story/lamda-sentient-ai-bias-google-blake-lemoine/), Jun 14. "Arguments over whether Google’s large language model has a soul distract from the real-world problems that plague artificial intelligence."
- Andre Sevenius Nilsen: [Artificial Consciousness Is Irrelevant - But Google’s LaMDA Is A Warning Sign](https://sevenius-nilsen.medium.com/artificial-consciousness-is-irrelevant-7dad40b77eed), Jun 16. This article redirects attention from the question of whether LaMDA is sentient to the question of how we humans should treat AI that passes the Turing test. Some [quotes from the article](https://hackmd.io/@alexhkurz/rJ21wGoF5).
- ...

This debate suggests that LLMs open up the possibility to explore a variety of philosophical questions (eg about consciousness) with methods of software engineering.

## Related
- David Chalmers: [How do you explain consciousness?](https://www.ted.com/talks/david_chalmers_how_do_you_explain_consciousness), TED 2014. [Articles (1993-2020)](http://consc.net/consciousness/).
- Jamie Woodhouse: [Humanism needs an upgrade: Is Sentientism the philosophy that could save the world?](https://sentientism.info/humanism-needs-an-upgrade-is-sentientism-the-philosophy-that-could-save-the-world#comment-493). "It may seem fanciful but surely to be future-proofed our ethics should help us think about how we should treat [general artificial intelligences] and how they should treat each other, even as we also worry about how they will treat us." May 21, 2020.
- William Buckner: [Symbolic reciprocity](https://traditionsofconflict.substack.com/p/symbolic-reciprocity), Jun 16, 2022. "One of the most important insights about people is that they can form social relationships with almost anything. Animals, the deceased, spirits, Gods, inanimate objects—humans look across nature and find new allies or adversaries everywhere they go."
- Moshe Vardi: [Ethics Washing in AI](https://www.youtube.com/watch?v=qW9WFOosaWc), June 23, 2022.
- Quanta Magazine: [By Exploring Virtual Worlds, AI Learns in New Ways](https://www.quantamagazine.org/ai-makes-strides-in-virtual-worlds-more-like-our-own-20220624/), June 24, 2022.
- Wikipedia: [Integrated Information Theory](https://en.wikipedia.org/wiki/Integrated_information_theory)

## Acknowledgments

Thanks to Jon Humphreys for providing some of the links.

[^elevators]: For more on sentient elevators making predictions about the future [see here](https://fb2bookfree.com/uploads/files/2021-12/1639865163_the-ultimate-hitchhikers-guide-to-the-galaxy.pdf#page=215).