| Transcript |
| 0:00 | · | We've seen a lot of headlines in the past year about how dangerous AI is and how overblown |
| 0:05 | · | these fears are . I've found it hard to make sense of this discussion. If only someone |
| 0:11 | · | could systematically interview experts and figure out what they're worried about. Well, |
| 0:16 | · | a group of researchers from the UK has done exactly that and just published |
| 0:21 | · | their results. What they have found is, not very reassuring. Let's have a look. |
| 0:27 | · | This new report is based on several rounds of interviews with 12 experts on software development |
| 0:32 | · | using what's called the Delphi method. The Delphi method is named after the Oracle of Delphi, |
| 0:38 | · | a position held by a priestess in the Greek city of Delphi around 2500 years ago. The |
| 0:44 | · | Oracle's task was to supposedly convey messages from the gods about the future. |
| 0:49 | · | The Delphi method was invented by the American non-profit RAND corporation in the 1950s to |
| 0:56 | · | make better use of experts' knowledge. It works by conducting in-depth interviews |
| 1:02 | · | with the experts. The interviews are then transcribed and anonymously shared with the |
| 1:06 | · | other participants. They add opinions on each other's interviews and further information, |
| 1:12 | · | then another round of interviews is done. This process can be repeated several times. |
| 1:18 | · | The Delphi method has become a common way for companies and committees to leverage expert |
| 1:24 | · | knowledge and convert it into actionable plans, and that's what these researchers also did. |
| 1:30 | · | They asked a lot of questions about what would happen in software development by the year 2040 |
| 1:36 | · | and eventually identified 5 points on which the experts more or less agreed. |
| 1:41 | · | The first one is that they all agree that by 2040 corners will be cut in AI safety. But |
| 1:47 | · | interestingly enough, they think it's not because of competition between companies, |
| 1:51 | · | but because of competition between nations, in particular they name the United States and China. |
| 1:58 | · | The results are summarized in this chart where blue means agreement, orange disagreement |
| 2:03 | · | and white means no opinion . Two of the experts said that by 2040 AI would cause |
| 2:09 | · | events with at least a million deaths, that's a megadeath. Yes, megadeath is actually a unit, |
| 2:16 | · | not just the name of a heavy metal band. You can also see that several experts disagree, |
| 2:21 | · | but this is partly because they think it will "only" be a few thousand fatalities. |
| 2:26 | · | Another thing on which the experts all agree is that by 2040 quantum computing |
| 2:31 | · | will only just be used. Again, you can see that some of them disagree, but in the text, |
| 2:35 | · | it's explained that they disagree by degree, in that one could say quantum computing is |
| 2:40 | · | already being used today it just has no commercial relevance and that's not going to change by 2040. |
| 2:47 | · | The next point of agreement is that almost all of them are worried that AI will make it increasingly |
| 2:53 | · | hard to tell apart truth from fiction in various domains from written text to image to video, |
| 3:00 | · | and that it will likely come to an arms race in which some Ais produce fake content and other Ais |
| 3:06 | · | will constantly try identify content as fake, quite possibly sometimes accidentally flagging |
| 3:12 | · | the truth as fake. It'll be a mess. One of the participants summarized it like this: |
| 3:18 | · | "We're not going to be living in a George Orwell world.… We're going to be living in a Philip K. |
| 3:23 | · | Dick world [where] nobody knows what's true." And just in case you're too young to remember, |
| 3:28 | · | Philip Dick wrote a bunch of dystopian future novels in which his characters |
| 3:34 | · | frequently question the nature of reality, the most famous probably being "Do Androids Dream |
| 3:40 | · | of Electric Sheep?" which was later adapted for the movie Bladerunner. |
| 3:44 | · | Now those three points I basically expected to see, but the last two I found somewhat of |
| 3:50 | · | a surprise. The experts all agree that by 2040 it will become common to buy and own |
| 3:56 | · | internet assets by way of tokenship. A tokenship is basically a digital record |
| 4:01 | · | and it's what NFTs have become known for. Even more interestingly, they don't think that this |
| 4:06 | · | tokenization will happen through blockchain technology but through other distributed |
| 4:12 | · | services. According to one of the interviewees "Blockchain has now proved its irrelevance." |
| 4:18 | · | And the final item is that they think the increasing complexity |
| 4:22 | · | of software in general and that of AI in particular will make it hard to tell |
| 4:26 | · | apart accidents from deliberate manipulation, basically because |
| 4:31 | · | no human will be able to really figure out what's going. Modern-day Kafka basically. |
| 4:37 | · | The experts also came up with a bunch of proposals for how to address these issues. |
| 4:43 | · | As you'd expect they ask for regulations on AI safety and more built-in safety requirements and |
| 4:49 | · | outcome checks on software development. This is what is listed here as "ambient |
| 4:55 | · | accountability". They also ask for better education of people in relevant positions |
| 5:00 | · | and more input from the social sciences on what the impact of all these changes might |
| 5:05 | · | be. These are surely all good ideas and they'll surely all be pretty much ignored. |
| 5:11 | · | I am confident these experts know what they're talking about, but I think they have somewhat of |
| 5:16 | · | a blind spot in an area that I care a lot about which is scientific publication. AI |
| 5:23 | · | is going to make it dramatically easier to produce rubbish papers and fake data |
| 5:28 | · | and spread these all over the globe. In fact, I would bet it's basically happening as we speak. |
| 5:34 | · | This falls into the general category of fake news and misinformation, |
| 5:39 | · | but I'd argue it's an underestimated special case. That's because fact checkers heavily |
| 5:44 | · | rely on scientific publication, and if that base erodes the entire house will tumble down. |
| 5:51 | · | So yes, interesting times ahead, maybe we'll soon find out whether androids do |
| 5:55 | · | dream of electric sheep and if they do, whether that makes them vegan. |
| 6:00 | · | By the way just wrote a new article for Nautilus magazine, |
| 6:05 | · | it's about Jonathan Oppenheim's new theory of post quantum gravity. |
| 6:10 | · | [ad text redacted] |
| 6:51 | · | Thanks for watching, see you tomorrow. |
