# tv   Shift  Deutsche Welle  January 25, 2020 7:15am-7:30am CET

7:15 am
our well coming up here on g.w. stay tuned for technology show shift that's up next and i'll forget in the meantime all the latest news of able or around the clock on our web site that's d.w. dot com or you can follow us on twitter at the news i'm going to ask him for lent back with more news at the top of the hour. literature invites us to see people in particular. like to see is the kids by. mike. brooks on youtube. in the. climate change. to make this. much mystery people.
7:16 am
want to do years do they have their future. g.w. dot com african megacities get. clear to. me in. this special edition of shift we explore how algorithms influence our lives. algorithms have been around for a long time but in the age of big data and digitalisation they're becoming ever more important and powerful are we letting algorithms think for us. what are algorithms. that's actually quite simple an algorithm is a clear process or set of rules to follow to solve a problem assembly instructions for example step by step they tell us what to do if
7:17 am
we follow them we should achieve the desired result assuming the instructions are good but that's another story. algorithm some probably existed for as long as people could think to convert sample the algorithm for computing square roots known as herons method was known to the babylonians and so mathematical algorithms were already known an ancient times one famous example is but. it states that in a right angled triangle the square of the hotness is equal to the sum of the squares of the other 2 sides in other words a squared plus b. squared equals c. squared. so if you know the length of 2 sides you can work out the length of the other. but where does the word algorithm come from it's named after me. people
7:18 am
how to perform written calculations using. these algorithms are still the ones we learned in school today. basically every computer program is an algorithm which tells the processor precisely how to utilize the coming stream of data. and now algorithms or even. research on machine. think society needs to regulate their implementation. what we should hope to achieve certain standards. that are free of discrimination. that is something that we as a society must be prepared to fight for. work on the internet.
7:19 am
facebook and government agencies and. otherwise they couldn't handle the huge streams of data but their eagerness to collect masses of data raises questions.
7:20 am
7:21 am
but catherine jar moules says algorithms aren't solely to blame because they're programmed and fed by humans algorithm learns because we train it using machine learning so either whether it's deep learning with a neural network so to speak or it's shallow machine learning where feeding it data and were asking it to form an opinion and of course when that opinion is based on data that has unfair treatment of groups then the algorithm or very much learned this like translation software with the millions of texts that teach it mention more female than male kindergarten teachers the algorithm learns and remembers that . picture recognition learns from large image databases here you find more pictures of women in a kitchen than men so the algorithm learns to associate women with kitchens
7:22 am
changing that is tough and time consuming. the coding is often top secret so it's difficult to prove an algorithm is prejudiced the theme is program can check for discrimination. it simulates fake accounts that are identical in everything but gender. accountability transparency and fairness are really the future of machine learning and we need to make them trustworthy and we need to you know allow people to ask questions of how they work and allow people to see how they're operating and how they're trained. to eliminate prejudice and algorithms it will take more social awareness and human intervention. for a long time it was considered an impossible task sequencing the wheat genome but now an international research team has done just that with the help of algorithms. that. over 200 researchers from 20 countries work together to discover bread wheats
7:23 am
d.n.a. in science magazine they reported how they were able to sequence over 100000 genes has 5 times more genes than humans so the researchers developed special algorithms to help them. while informative. not all took part in the project. and lets the on in recent years new algorithms were developed specially to piece together the different fragments of the genomes and they really helped us to create a coherent sequence out of this huge hile of puzzle pieces of printing the researchers hope this will help them break new week for ideas that are more resistant to climate change. and possibly have to make it more adaptable to different climatic conditions for example longer periods of drought and possibly longer periods of heavy rain as well. along with the rice we disconcert of the world's most important foodstuffs experts warn that we must increase if
7:24 am
we're to satisfy the hunger of earth's growing population. the sequencing of the bread wheat genome is a major step in that direction. can algorithms predict the next best seller. a startup hopes to use its software to help publishers fish out the most promising manuscripts in a matter of seconds. what novels will fly off the bookstore shelves and which ones won't. a piece of software named qualification is supposed to be able to predict which manuscript will become the next best seller. the idea came from hearing from many publishers and authors too that much of a huge amount of material that submitted each year in germany doesn't even get read . simply because it would take too many people to do so.
7:25 am
the ai software rates stories according to their sales potential using 100000 books and their sales figures the program learned to recognize a bestseller sentence by sentence test readers fed it with successful novels and rated them. the criteria including sentence length suspense book happy larry and emotions. didn't even take a books and divide it into its literary components as we explore what kind of themes the work uses on what is its dramatic arc what style of language or a constellation of characters is employed going from home and using all these characteristics we make a final rating i know. several publishers are currently testing the software including small hamburg based publisher. its director and one editor regularly wade through mountains of manuscripts. say we get sent 100 manuscripts you can be pretty sure that a lot of these can be filtered out because they're not suitable for
7:26 am
a formal reason i'll say. if the software manages to filter out that 10 to 20 percent and that would be a great help. but after that separating the wheat from the chaff is really all about the little details. ror says people will always be needed to work closely with the authors. many in the publishing business are critical of the software including kissed and primus who's worked as a book editor for 25 years. then. if lots of publishers start using it i fear there will be more mass market literature released more things that cater to the mass market. i must not under that's happening already but i'm afraid it will only get worse and that really innovative different and surprising works will fall by the wayside. also the founder say
7:27 am
fears that the book market could become less diversified are unfounded. beguiled and yet i'm getting tired especially think the opposite is true it will likely lead to more diversity because publishers have the opportunity to assess whether something might really succeed as it contains many elements that are really new and work well sent on quote false and. christiane good into moonlight says a writer of dark fantasy novels and sees the software as an opportunity for authors like yourself. it could help him learn how to better reach his target audience and tell him if it's worth investing in marketing. who just sent his latest work to quali fiction for analysis. business one so listen to see it's a bit darker than normal. this is the average just you're a bit more intense. here you can see that you manage to maintain the suspense but over time you get more and more extreme. and i know. in almost every respect you
7:28 am
meet the target groups expectations of. the algorithm thinks christan going to his novel has a 23 percent chance of becoming a bestseller. this is the sort of test and that's just the stats and you need to look at it from the view of a publisher looking to appeal to a mass audience and and for that it's a pretty good rating yet no question this was not bad for an unknown writer. of fiction was designed to help publishers who lack the time to read all of the manuscripts they received but authors also appreciate the software speed back for all its apparent prejudices at least we can say that ai doesn't judge a book by its cover.
7:29 am
hi there. on the audi r s q 8. don't i look great. ok ok so i'm not what you call danger and i'm a bit of a gas guzzler son supersport in fact just. today someone who isn't a big fan of s.u.v.s wants to test me out. to see if i can win them over red. next on the c.w. . every 2 seconds the person is forced to flee their home. the consequences been disastrous for our documentary series displaced to fix dramatic humanitarian crises for a problem world. forgetting we don't have time to think i didn't go to university to kill people but accidentally i mean
7:30 am
a handful of people feel for their lives and their future so they seek refuge abroad but what will become of summers who stay behind. displaced this week on d w. live. live . lists. lists. we are living during the most extraordinary time mr.

10 Views