Prime Time 7/8, Writing, Arbeitsheft

16 66 16 Science and new technologies 16.1 Text input Artificial intelligence isn’t as clever as we think, but that doesn’t stop it being a threat A new report tries to bring order to the messy business of measuring AI progress. How clever is artificial intelligence really? And how fast is it progressing? These are questions that keep politicians, economists and AI researchers up at night. And answering them is crucial, not just to improve public understanding, but to help societies and governments figure out how to react to this technology in the coming years. The AI Index, a recent report from experts at MIT, Stanford University, OpenAI and other institutions, focuses on two main points: first, that the field of AI is more active than ever before, with minds and money pouring in at an incredible rate; and second, that although AI has overtaken humanity when it comes to performing a few very specific tasks, it’s still extremely limited in terms of general intelligence. Even though many people overestimate the capacity of AI, there are definitely tasks where AI has already matched or eclipsed human performance. These include identifying common objects in images (on a test database humans get a five per cent error rate; machines, three per cent), and transcribing speech (as of 2017, a number of AI systems can transcribe audio with the same word error rate as a human). A number of games have also been definitively conquered, including Jeopardy, Pac-Man and, most famously, Go. But as the report says, these metrics only give us a partial view of machine intelligence. For example, AI may be just as accurate as humans when it comes to writing down recorded dialogue, but it can’t gauge sarcasm, identify jokes or account for a million other pieces of cultural context that are crucial to understanding even the most casual conversation. The AI Index acknowledges this and adds that a bigger problem here is that we don’t even have a good way to measure this sort of common-sense understanding. There is no IQ test for computers, despite what some PR people claim. Does that mean we need to worry less about AI’s effects on society? Unfortunately, not. Even though our most advanced AI systems are dumber than assumed, it won’t stop them from having a huge effect on our lives – especially in the world of work. A recent study published by consultancy firm McKinsey suggested that as many as 800 million jobs around the world could be under threat from automation in the next twelve years. But the study’s authors clarify that only six per cent of the most rote and repetitive jobs are at a risk of being automated entirely. For the rest, only parts of the job can be done by machines. This is where the narrow intelligence of AI will really have an impact, and here it’s tricky to say what the effect will be. If a computer can do one-third of your job, what happens next? Do you get trained to take on new tasks, or does your boss fire you, or some of your colleagues? What if you just get a pay cut instead? Do you have the money to retrain, or will you be forced to take the hit in living standards? It’s easy to see that finding answers to these questions is incredibly challenging. And it mirrors the difficulties we have understanding other complex threats from artificial intelligence. For example, while we don’t need to worry about super-intelligent AI running amok any time soon, we do need to think about how machine learning algorithms used in healthcare, education and criminal justice are making biased judgements. The conclusion of both the AI Index and McKinsey’s study is that these questions, and others, need deep consideration in order to stay ahead of what’s coming. As machines get clever, we can’t afford to be dumb. (James Vincent, www.theverge.com , 1 December 2017) 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Nur zu Prüfzwecken – Eigentum des Verlags öbv