(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2023 Retraice, Inc.)

Re103: Dimensions of Intelligent Agents
(Day 1, AIMA4e Chpt. 1)

retraice.com

Adding fitness and truth to AIMA's dimensions, and considering `artificial flight'.
Deciding vs. discovering what intelligence and AI are; Vico on knowledge and engineering; the spectrum view and the category view of intelligent agents; `artificial flight', birds and intelligence.

Air date: Sunday, 1st Jan. 2023, 10:00 PM Eastern/US.

Amendments and corrections: Re102

Kurzweil doesn't seem to ever have said the Singularity will happen in 2042, as I mistakenly recalled. See Retraice (2022/12/31) for details. Putting something on the calendar is still worth doing, so why not 2042?

On expert predictions found lacking, see Russell & Norvig (2020) p. 28 on Tetlock (2017).

Deciding vs. discovering what AI is

Russell & Norvig (2020) (p. 1) describe the contents of Chapter 1 as, in part, being about deciding what AI is. Is this a subtle error? Does it have effects on our understanding? Shouldn't we be at least as focused on discovering what AI is, or at least what intelligence is? It highlights a core difficulty in AI anticipated by the Italian philosopher Giambattista Vico (1778-1744): "one is certain of only what one builds"^1 or "the true and the made are convertible"^2

Agent intelligence with respect to truth and fitness
________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

PIC
Agents might be in a space, along three measures of degree, or in one of eight categories. Based on the Russell & Norvig (2020) categorization (pp. 1-5) and Hoffman (2019). On our `RTFM' model of the `good', see Retraice (2022/10/24).
________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________

`Artificial flight'

Russell & Norvig (2020) argue that `the Turing test approach' to AI is inadequate by analogy: We didn't succeed in building planes until we gave up imitating birds.

What is the connection to AI safety? Imagine a world with `artificial birds' flying around. What would make them concerning? Their intelligence, and attendant motivations and capabilities. The dangers of accidentally loosing artificial birds would be the same as those of artificial intelligence: the intelligence and agency. We want Star Wars, not the Matrix; we want R2-D2, not Agent Smith.

On swimming and submarines, see Retraice (2022/11/23), Retraice (2022/12/26). On Cornelis Drebbel, who invented the thermostat and the submarine, see Russell & Norvig (2020) p. 16. On Dennett and `colleagues' vs. `tools', see Retraice (2022/11/16) and Dennett (2019).

_

References

Brockman, J. (Ed.) (2019). Possible Minds: Twenty-Five Ways of Looking at AI. Penguin. ISBN: 978-0525557999. Searches:
https://www.amazon.com/s?k=978-0525557999
https://www.google.com/search?q=isbn+978-0525557999
https://lccn.loc.gov/2018032888

Dennett, D. C. (2019). What can we do? (pp. 41-53). In Brockman (2019).

Genesereth, M. R., & Nilsson, N. J. (1987). Logical Foundations of Artificial Intelligence. Morgan Kaufmann. ISBN: 0934613311. Searches:
https://www.amazon.com/s?k=0934613311
https://www.google.com/search?q=isbn+0934613311
https://lccn.loc.gov/87005461

Hoffman, D. (2019). The Case Against Reality: Why Evolution Hid the Truth from Our Eyes. W. W. Norton & Company. ISBN: 978-0393254693. Searches:
https://www.amazon.com/s?k=978-0393254693
https://www.google.com/search?q=isbn+978-0393254693
https://lccn.loc.gov/2019006962

Honderich, T. (Ed.) (2005). Oxford Companion to Philosophy. Oxford University Press, 2nd ed. ISBN: 0199264791. Searches:
https://www.amazon.com/s?k=0199264791
https://www.google.com/search?q=isbn+0199264791
https://lccn.loc.gov/2005275452

Retraice (2022/10/24). Re28: What's Good? RTFM. retraice.com.
https://www.retraice.com/segments/re28 Retrieved 25th Oct. 2022.

Retraice (2022/11/16). Re52: Big Questions About AI. retraice.com.
https://www.retraice.com/segments/re52 Retrieved 17th Nov. 2022.

Retraice (2022/11/23). Re59: Learning, Interacting, Conclusions (AIMA4e chpts. 19-28). retraice.com.
https://www.retraice.com/segments/re59 Retrieved 24th Nov. 2022.

Retraice (2022/12/26). Re96: News of ChatGPT, Part 1. retraice.com.
https://www.retraice.com/segments/re96 Retrieved 27th Dec. 2022.

Retraice (2022/12/31). Re102: AI For What. retraice.com.
https://www.retraice.com/segments/re102 Retrieved 1st Jan. 2023.

Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach. Pearson, 4th ed. ISBN: 978-0134610993. Searches:
https://www.amazon.com/s?k=978-0134610993
https://www.google.com/search?q=isbn+978-0134610993
https://lccn.loc.gov/2019047498

Footnotes

^1 Genesereth & Nilsson (1987) p. 1

^2 Honderich (2005) p. 945.