Scribd
Upload
21 views

AI Values

Uploaded by

Adarsh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
21 views

AI Values

Uploaded by

Adarsh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 7
Al VALUES (ETHICAL DECISION MAKING) Learning Outéomesiinm: s around Al * Aland Ethical Concems What do we do about the Biases in AI? '@ product of human imagination. Prejudice or bias, whether we are willing to admit nan experience. In order to avoid prejudice, programmers can tur to the philosophical basis of curiosity, seeking to transcend our own perspectives, and learning to see the world outside 1 human algorithms. Machines are everywhere and they lear from humans. Since all humans e:nies and biases based on the environment and cultures they are surrounded by, so, the machines also ses from humans. movies on YouTube and answer the questions: is a problem: racist soap dispenser’ at Facebook office does not work for black people: pologizes for Tagging Black Couple as Gorilas: ould be the reason for bias? ical Decision Making) | 187 —— “ay values (Ethi > = — vs an unconscious technol § nye ye com oss took nny Hi, after all tne ypany behi yo eae onthe surface, small mistakes! f ‘ret sce in fac, they show why diversi) wil ‘nas nat being aetiveratel 2° os voten pred arve owever mectors by othe link aoc nent sean the QR code orgoto ine eaten Below: Humans Need NOt AF per S e more | is video? gre more What you learned from t a pris syste control Of benefit a py learnit ntellige Issues and Concerns around AI trust R ficial intelligence, the Problems and concerns that orl ia en artificial intellige ns tificial intelligence. out sore of the esues andl cOncEET: surrounding artifi 95 fc5 z cognitiv biases © Job Loss Problem waren’ the ope Major rut Safety Problem Computing Power ‘Al uses a lot of unstructured data. Unstructured data is of great value to enterprises, but m a eras ee they cannot use traditional systems for analysis eee : ‘agement system (RDBMS), which makes it eee yypes include audio and video files, images, text, web content, etc. If thro Pi processed and stored as easier-to-understand analytical data, it will be easier f0 better logical decisions, Obtaining and fun th level of computing power can be 's Obtaining and \g and funding such level of computir becnaa 188 | Touchpad Arica! intetigence-xi wen Hficial Intellige I problem oe Fe pectation from Al today. The mea ue Orr obs. So, the IT Indust Says, Arti ST ee hag nn npn oer fe theatening work. ft wi act £2" and defintely curtly convey that Ais ust a optimiza 0st automation of routine, lo 9 N Of industrial work and better data- technology becomes, the more it car : In be used fo sik, cloud servi ‘i IF nefariou re ihe Oe social platform, hackats vil take advantage of sy sanction nthe ey hav portunity. Regardless of the hacker's motivati age of any vulnerabilities in the ‘d business information is always a loss because this Retain cbs ied x ses can for the hackers’ roblem: algorithms become more powerful, it also brings up some issues related to confidence ir decisions that will benefit humanity, As [Artificial Intelligence slowly reaches human-level the issue of trust becomes more important. All stakeholders must bbe aware of the inherent bias detection mechanism and processing methods. «key factor that plays an important role in reducing the trust ts capabilities, and even the defects related to Al. \s, The Al algorithms must have a strong ‘gap. The user must understand > Video Session 2 OR code or go to the link to wal +: Al. Is Monitoring You Right Now hove KpybityiXts here the computer forecasting systems collect te “hea ‘wing game on Kahoot: re/thought-€xP Brainy Fact ched by Microsoft on as tga i iri hnes cuter tet neh Foe posts Oa Bee err rier on watch 2 2090 0" cyused » STUY oe Ra ‘and offensive tweets through its Twat ee oe hours after its launc! ied base prompting Microsoft to shut down te serViCeI' | cg asthe bot respondet used by trolls that “att Microsoft, this was ca on its interaction with people on Twit! Al and Ethical Concerns Js with human moral obliga ions and duties. It is a set of als with human icial Intelligence ethics is a py individuals or groups. Artificial ; 1 indo a achines. Various Al ethical concems t also take human-like decisions? | the Al algorithms be fair and ju ie Ethics is defined as the dscipin tat de actions of that regulate the behaviour and rtificially intelligen' technology specific to robots and other a + If artificial intelligence produces human-like results, can ! will ‘thy of human confidence? fe] + data-based training + prediction. + If artificial intelligence decides whether to grant bank loans: + Ifartifcial intelligence makes human decisions, is it also wor + Artifical intelligence is basically data + mathematical mod! provided for training is unintentionally skewed or biased? * Will Altake over jobs? These questions and concerns come up all the time, so the "ethics of Artificial Intelligence” is very im must understand how artificial intelligence works and the framework of artificial intelligence ethies, P Video Session Scan the QR code or go to the link to watch the video ar i ind answer the qu below: Al Ethics: Why It Matters? baa https//+wwen.youtube.com/watch?v=C2MWrCdGGOQ What have you learned from this video? Explain in minimum four lines 5 @a and Bias ,eof the biggest challenges facing | “ble bias when exploring the depth ioe a problem with AI systems j at : IS that cial, gender, community, of ae fair ci a ae these biases may intensify nae ea io ea) Makes im pe herefore, the immediate den —— terpret. pear te o tysteme vin, ie al ‘will continue to use rect da eee nn wr ye the sources of AI bias? iy three sources of Al bias that are: t ceo «s good as the data we feed them, And inputting skewed data into the system leads to AL bias. AL derstand whether their training datas right or wrong and represents a broader base, eloped an Al tool for recruitment, but the company 509° realized its new system was ‘gender-biased. computer models were trained to inspect applications by observing patterns inresumes submitted to the over a 10-year period. Most resumes were submitted by men. ‘50 the Al system started ‘favouring males obs in the company. > Video Session can the QR code or visit the following link ven below: Amazon's Sexist Recruitment. parenting, experience, aj ye case. is not th ‘Alis the sam f world around eher. t's Naa Para unbiased Unlptions about the Wi" wthink” the way itis taught, Technology is supposed 10 Be Mh assumptions mg it tends nize people people and therefore inter these peoP 1u upload your selfies here, erent or a arow our portal Pea webate poral ort generato" ("PSA pangisg spore dn mgr 1 Bard. The drawback (5 se oe eae Mond to make you look satabase whe J intolerance or institutional discri on ssn ay tha le eal reinforce societal bases. | eee ing. Google recently found itself in ti bias in A-based decision-ma it fuding owners or employers, to disci ‘ ers, including © . ertiser i pa je-owned YouTube can choose to exclude sn Google oF 6009 a ntentonally or unintentional irae Ts erates the anti-discrimination law. Since th male or female, whic igh technology =) What do we do about the Biases in AI? 1 underlying human biases. With more advanced tools for detecting = for humans. This can take the form of running an algorithm, and then a esults. The human decision maker can also use “interpretation techniques* «2 to the models decision, to understand the reasons for possible discrepancies. 1 itis not enough to change the algorithm; business leaders must. ses behind it chines can work together to reduce bias, Some "human-in-the- fe options for humans lations or prov 3 {0 choose from. The transparency of these al ‘and how much weight they should be given, ore, provide more and take a multidisci y of course ath 0 make designer choices more transpore sig “more transparent and incorporate ethics can help people to under Companies should invest mo data pee a, plinary approach to bi Progress in this area. Major efforts t '™o computer science courses are a Touchpad Articiy Intelligence ., E ployment and AI 7) ¢ tific ial intelligence has vastly 4 iy it of ALon employment and nena on jobs at Risk «< and repetitive jobs wil i jobs will be easily automated. This will gradually lead to cerain jobs becoming rae ae acta a ae to nee ‘opening new bank accounts, and ‘other customer So, the intial queries may be handl ries will be handled by customer car Ea noe eros ¢ y cu xe representatives. This wll also Pring ‘down the load on the epresentatives as most initial queries tend to be the same, cles will soon be a reality cess large amounts of data like financial services, insurance may also be affected to @ certain ay a significant role in iminating red tape thereby improving various services to citizens impact the healthcare sector. With better image cdassifiers, the intial diagnos of diseases will ‘¢ load and lead to timely treatment. proficient at Natural Language Processing. Henct simple transations may be handled PY Aiwnile iranslators can concentrate OP he complicated translations: e {ew jobs obsolete, Dut also created new jobs. For advent of computers and later the internet made bu ren | Js reduced the number of cashiers IN tpanks but it had 2n a positive impact bpanking services: ctiegire ghee rer SI Arvigo ereate more jobs tan te es that tl alec TMe S professions that Al will disrupt

You might also like