AI’s fu­ture is not so scary

Jamaica Gleaner - - SPORTS - – MIT Tech­nol­ogy

AR­TI­FI­CIAL IN­TEL­LI­GENCE (AI) will trans­form just about ev­ery­thing, but tech­nol­o­gists should stop fret­ting that it’s go­ing to de­stroy the world like Skynet.

The odds that AI will en­slave or elim­i­nate hu­mankind within the next decade or so are thank­fully slim. At the same time, how­ever, AI looks cer­tain to up­end huge as­pects of ev­ery­day life, from em­ploy­ment and ed­u­ca­tion to trans­porta­tion and en­ter­tain­ment. So con­cludes a ma­jor re­port from Stan­ford Univer­sity, co-au­thored by more than 20 lead­ers in the fields of AI, com­puter science, and ro­bot­ics. The anal­y­sis is sig­nif­i­cant be­cause pub­lic alarm over the im­pact of AI threat­ens to shape pub­lic pol­icy and cor­po­rate de­ci­sions.

The re­port pre­dicts that au­to­mated trucks, fly­ing ve­hi­cles, and per­sonal ro­bots will be com­mon­place by 2030, but it cau­tions that re­main­ing tech­ni­cal ob­sta­cles will limit them to cer­tain niches. It also warns that the so­cial and eth­i­cal im­pli­ca­tions of ad­vances in AI, such as the po­ten­tial for un­em­ploy­ment in cer­tain ar­eas and likely ero­sions of pri­vacy driven by new forms of sur­veil­lance, will need to be open to dis­cus­sion and de­bate.

The study, part of a project in­tended to last 100 years, is some­thing of a re­but­tal to some of the alarmist pro­nounce­ments that have been made about AI. “No ma­chines with self-sus­tain­ing long-term goals and in­tent have been de­vel­oped, nor are they likely to be de­vel­oped in the near fu­ture,” the re­port says.

“I re­ally see this as a com­ing-ofage mo­ment for the field,” says Oren Etzioni, CEO of the Allen In­sti­tute for Ar­ti­fi­cial In­tel­li­gence, an in­de­pen­dent re­search in­sti­tute in Seat­tle, who is a co-au­thor of the re­port. “The ex­treme pos­i­tive hype is wrong, and the fear-mon­ger­ing is not based on any data.”

The re­port iden­ti­fies the most promis­ing ar­eas for fu­ture AI re­search, and Etzioni says key among these is re­search on ways for hu­mans and AI sys­tems to col­lab­o­rate ef­fec­tively. Stan­ford’s one hun­dred year study on ar­ti­fi­cial in­tel­li­gence will re­port find­ings ev­ery five years. The first re­port fo­cuses on ar­eas in which AI will have a sig­nif­i­cant im­pact, in­clud­ing trans­porta­tion, health care, ed­u­ca­tion, and em­ploy­ment.

Newspapers in English

Newspapers from Jamaica

© PressReader. All rights reserved.