Eenglish-tests.com

C2 Text 61 · Symposium · 717 words

The Singularity Debate

A computer scientist and a philosopher debate whether machines will one day outthink their makers, and find they agree more than expected.

The following is an edited transcript of the closing session of this year's Ludlow Festival of Ideas, in which a computer scientist and a philosopher were asked whether machines will one day outthink the people who build them. The session was chaired by the broadcaster Imogen Hartley.

Imogen Hartley: Let's start with the word itself. 'Singularity' sounds rather grand. Where does it come from? Dr Rafael Osei: It's borrowed from mathematics and physics, where a singularity is a point at which the usual equations break down. The mathematician Stanislaw Ulam recalled a conversation in the 1950s with John von Neumann about technological progress approaching some such point. Then, in 1965, the statistician I. J. Good described an 'ultraintelligent machine' that could design even better machines, setting off what he called an 'intelligence explosion'. He added, with a certain dry wit, that such a machine would be the last invention we ever needed to make. Imogen Hartley: And the science-fiction writer Vernor Vinge popularised the term in an essay in 1993. Dr Rafael Osei: Exactly. The idea's pedigree is respectable. My quarrel is with what it has become: less a hypothesis than a horizon, always thirty years away.

Professor Miriam Calloway: I'd put it slightly differently. I don't think the singularity is a prediction at all; it's a thought experiment that has been mistaken for a forecast. Its value lies in forcing us to ask what we would do if a system could improve itself faster than we could understand it. You needn't believe it will happen in order to take that question seriously. Imogen Hartley: Rafael, you're sceptical about the timescale. Is that because the technology is slower than people claim? Dr Rafael Osei: Partly. Progress in machine learning over the past decade has been genuinely startling, and I say that as someone who used to scoff at it. But the argument for an explosion rests on the assumption that intelligence is a single quantity, like the volume on a radio, which you can simply turn up. I'm not convinced it is. A system can be superb at one task and hopeless at the next. Each improvement may get harder, not easier: the law of diminishing returns applies to cleverness as much as to anything else.

Professor Miriam Calloway: That's a fair point, but it cuts both ways. If intelligence isn't one thing, then 'human-level' isn't a single finish line either, and machines may overtake us in a patchwork fashion, area by area, without any dramatic moment we could point to. The danger isn't a thunderclap; it's a slow erosion of our ability to check what our tools are doing. Imogen Hartley: So the singularity might arrive without anyone noticing? Professor Miriam Calloway: Or it might never arrive and still have been worth worrying about. Insurance is not a waste of money simply because your house didn't burn down.

Imogen Hartley: A question from the audience, from a sixth-former, I'm told. 'If we can't agree what intelligence is, how will we know when a machine has more of it than us?' Dr Rafael Osei: That's a better question than most of the ones I get at conferences. Honestly, we might not. Tests that seemed definitive twenty years ago, such as beating a grandmaster at chess, turned out to measure something narrower than we had assumed. Every time a machine passes a test, we tend to move the goalposts and declare that the test wasn't about real intelligence after all. Professor Miriam Calloway: Which is either a sign of human vanity or of genuine conceptual progress. I suspect it's a bit of both.

Imogen Hartley: We're nearly out of time. Is there anything you agree on? Dr Rafael Osei: More than the programme suggests. We both think the breathless headlines do real harm. They make people either panic or switch off, and neither helps. Professor Miriam Calloway: And we both think the interesting decisions are being made now, not at some imagined moment of rupture: who builds these systems, who checks them, and whose values are written into them. If there is a singularity, it won't be something that simply happens to us. It will be the sum of many small choices we have already made. Imogen Hartley: On that sobering but oddly hopeful note, thank you both.

Words in colour are explained in the glossary below.

Glossary

WordMeaning
dry wit/ˌdraɪ ˈwɪt/a quiet, clever kind of humour, said with a serious face
popularised/ˈpɒpjələraɪzd/made something known and liked by many people
pedigree/ˈpedɪɡriː/the history and origins of something, which make it respected
thought experiment/ˈθɔːt ɪkˌsperɪmənt/an imagined situation used to explore an idea, not a real test
scoff at/ˈskɒf æt/laugh at something and refuse to take it seriously
diminishing returns/dɪˌmɪnɪʃɪŋ rɪˈtɜːnz/the idea that each extra effort brings a smaller gain than the one before
cuts both ways/ˌkʌts bəʊθ ˈweɪz/can be used to support both sides of an argument
patchwork/ˈpætʃwɜːk/made of many separate, uneven pieces
erosion/ɪˈrəʊʒən/a slow wearing away or weakening
definitive/dɪˈfɪnətɪv/final and certain; settling a question once and for all
move the goalposts/ˌmuːv ðə ˈɡəʊlpəʊsts/change the rules or the standard of success after someone has met it
vanity/ˈvænəti/too much pride in yourself and how special you are
breathless/ˈbreθləs/(of news) excited and exaggerated
rupture/ˈrʌptʃə/a sudden, complete break
sobering/ˈsəʊbərɪŋ/making you feel serious and thoughtful

The Singularity Debate: tests

Found this text useful? Share it with a friend or your class.