Largest open collection quotes about AI

Link post

I apol­o­gize for my bad English, this is not my na­tive lan­guage. And prob­a­bly I will make some mis­takes when post­ing.

For over 2 years I have been read­ing ma­te­ri­als on the topic of AI Safety. I don’t have the ap­pro­pri­ate ed­u­ca­tion, cog­ni­tive abil­ities, knowl­edge. I do not even have time to learn the lan­guage. So I didn’t hope to do some­thing use­ful my­self.

But once I tried to sys­tem­atize quo­ta­tions from one show in or­der to un­der­stand when the ex­perts rep­re­sented there are wait­ing for AGI and how likely they con­sider the ex­tinc­tion of hu­man­ity.

I thought it would be in­ter­est­ing to do so with the rest of the ex­perts.
In ad­di­tion, I have already seen and stud­ied with in­ter­est such col­lec­tions of quotes. It seemed to me that the best thing I could do was try to do some­thing similar.

There­fore, I be­gan to col­lect quotes from peo­ple who can be at­tributed to the ex­perts. It turned out to be harder than I thought.

I have com­piled a table with quotes from more than 800 ex­perts. I tried not to dis­tort the opinion of fore­cast­ers and sim­ply copied from sources, some­times delet­ing or slightly edit­ing. My ed­its can be rec­og­nized by square brack­ets :)

1) The first column of the table is the name of the ex­pert.
2) The sec­ond column is the year of the fore­cast. The table is built in chronolog­i­cal or­der.
3) The third column is the pre­dicted time for AGI. Un­for­tu­nately, most peo­ple did not speak di­rectly about time and prob­a­bil­ity. Be­cause of this, many quotes came out rather vague. For ex­am­ple, “Machines are very far from be­ing in­tel­li­gent” or “And we can reach it in a close time”.
4) The fourth column is an opinion about Take­off Speed. About how much progress will be ac­cel­er­ated af­ter cre­ate of AGI.
5) The fifth column is the ex­pert’s opinion about the fu­ture of mankind with AGI. Choos­ing a quote here was the hard­est. Most of all I was in­ter­ested in the risk of ex­tinc­tion or se­ri­ous shocks due to AI, and I tried to provide quotes that most fully re­veal this par­tic­u­lar topic.
6) The sixth column in­di­cates the source of the quote.

That is, to the right of the fore­caster’s name, you can find out the date of the given quotes, his opinion about the time of the cre­ation of AI, about the in­tel­lec­tual ex­plo­sion and about the fu­ture of hu­man­ity, as well as get ac­quainted with the source.

Of course, cases where the ex­pert spoke on the topic of time, the speed of self-im­prove­ment and the in­fluence of AI in the frame­work of one ma­te­rial are quite rare. There­fore many cells are left empty.

I had to give sev­eral quotes per per­son, some­times they were sep­a­rated for years and even decades.

Since all the quotes are given in chronolog­i­cal or­der, the opinions of some peo­ple are “scat­tered” in the table.

For ex­am­ple, Gw­ern spoke about the fu­ture of mankind in 2010, about the growth of AI in 2014 and about the fore­casts for the emer­gence of AI in 2018.
How­ever, you can sim­ply use search.

In ad­di­tion, some­times one per­son has already made a cer­tain fore­cast but later changed or ex­panded his opinion. I tried to take into ac­count such quotes.

I also re­viewed anony­mous ex­pert in­ter­views and in­di­cated them. If the gen­eral list of re­spon­dents was known, I cited them as well.

It was difficult to de­cide who should be con­sid­ered an ex­pert and what quotes should be in­cluded in the work.

I had to make con­tro­ver­sial de­ci­sions. The table in­cludes a lot of peo­ple who are en­trepreneurs but may have in­sights on ad­vanced re­search. There are sev­eral fu­tur­ists and philoso­phers in the table. There are writ­ers like Clark and Vinge, whose opinion seems im­por­tant to me.

I have a ver­sion of this work with­out chronolog­i­cal sep­a­ra­tion, where the quotes are more grouped by name. Per­haps some­one will be more con­ve­nient.

It is difficult to draw con­clu­sions from the work. The ab­solute ma­jor­ity of ex­perts did not talk about ex­act dates and did not in­di­cate the prob­a­bil­ity of their pre­dic­tions.

I can only say that most fore­cast­ers do not ex­pect AI in the near fu­ture, do not ex­pect IE and seem op­ti­mistic.

In ad­di­tion, it seemed to me that in the twen­tieth cen­tury the lead­ing ex­perts were on av­er­age more pes­simistic: Tur­ing, Wiener, I. J. Good, Fred­kin, Shan­non, Mo­ravec, etc.

Young re­searchers are on av­er­age more op­ti­mistic than older ones—even in the field of AI Safety, where on av­er­age there are nat­u­rally more con­cerned peo­ple.

I think that to con­firm al­most any views you can find the opinion of a re­spected ex­pert.

I re­ally hope that for some­one my work will be use­ful and in­ter­est­ing. Crit­i­cism and ad­di­tions are wel­come.