Smarter ma­chines, smarter hu­mans

New Straits Times - - Bots - AFP

Raib­ert with the SpotMini robot.

TECH­NOL­OGY MEET­ING PEO­PLE’S NEEDS

“As ma­chines get smarter, so do we,” Gru­ber said.

“Ar­ti­fi­cial in­tel­li­gence can en­able part­ner­ships where each hu­man on the team is do­ing what they do best,” he added.

Gru­ber, a co-cre­ator of Siri and ar­ti­fi­cial in­tel­li­gence re­search at Ap­ple, told of be­ing drawn to the field three decades ago by the po­ten­tial for tech­nol­ogy to meet peo­ple’s needs.

“I am happy to see that the idea of an in­tel­li­gent per­sonal as­sis­tant is main­stream,” he said.

Now he has taken his in­no­va­tive ap­proach to smart ma­chines and is turn­ing the think­ing about the tech­nol­ogy on its head.

“In­stead of ask­ing how smart we can make our ma­chines, let’s ask how smart our ma­chines can make us,” Gru­ber said.

Al­ready smart tech­nol­ogy is tak­ing hold, with pop­u­lar dig­i­tal as­sis­tants like Ap­ple’s Siri.

South Korean gi­ant Sam­sung cre­ated Bixby to break into a surg­ing mar­ket for voice-ac­ti­vated vir­tual as­sis­tants, which in­cludes Ama­zon’s Alexa, Google’s As­sis­tant and Mi­crosoft’s Cor­tana.

Ama­zon ap­pears to have im­pacted the sec­tor the most with its con­nected speak­ers us­ing Alexa. The ser­vice al­lows users a wide range of voice in­ter­ac­tions for mu­sic, news, pur­chases and con­nects with smart home de­vices. RE­MEM­BER­ING EV­ERY­THING

Gru­ber en­vi­sions ar­ti­fi­cial in­tel­li­gence (AI) get­ting even more per­sonal, per­haps aug­ment­ing hu­man mem­ory.

“Hu­man mem­ory is fa­mously flawed, like where did the 1960s go and can I go there too?” Gru­ber quipped.

He spoke of a fu­ture in which ar­ti­fi­cial in­tel­li­gence re­mem­bers ev­ery­one met dur­ing a life­time and de­tails of ev­ery­thing some­one read, heard, said or did.

“From the tini­est clue it could help you re­trieve any­thing you’ve seen or heard be­fore,” he said.

“I be­lieve AI will make per­sonal mem­ory en­hance­ment a re­al­ity; it’s in­evitable.”

Such mem­o­ries would need to be pri­vate, with peo­ple choos­ing what to keep and be kept ab­so­lutely se­cure, he main­tained.

SUREFOOTED RO­BOTS

Bos­ton Dy­nam­ics ro­bot­ics com­pany founder Marc Raib­ert was at the re­cent 2017 TED Con­fer­ence in Van­cou­ver, Canada, with a four-legged SpotMini robot nim­ble enough to frolic amid the con­fer­ence crowd.

He smiled but would not com­ment when asked about the po­ten­tial to im­bue the gad­get with the kind of ar­ti­fi­cial in­tel­li­gence.

Raib­ert did, how­ever, note that the ro­bots are de­signed to be com­pat­i­ble with new “user in­ter­faces”.

Cur­rent vir­tual as­sis­tants have been de­scribed as a step into an era of con­trol­ling com­put­ers by speak­ing in­stead of typ­ing or tap­ping screens.

“It won’t be too long be­fore we’re us­ing ro­bots to take care of our par­ents or help our chil­dren take care of us,” Raib­ert said.

THE “GO­RILLA PROB­LEM”

How­ever, not ev­ery­one has em­braced the idea of a fu­ture in which ma­chines are smarter and more ca­pable than hu­mans.

Stu­art Rus­sell, a com­puter sci­ences pro­fes­sor, re­ferred to the sit­u­a­tion as the “go­rilla prob­lem” in that when smarter hu­mans came along, it boded ill for evo­lu­tion­ary an­ces­tors.

“This queasy feel­ing that making some­thing smarter than your own species is not a good idea,” said Rus­sell, co-author of the book Ar­ti­fi­cial In­tel­li­gence: A Mod­ern Ap­proach.

As an AI re­searcher, he sup­ported re­search in the tech­nol­ogy.

How­ever, he is all for pro­gram­ming ma­chines with robotic laws of be­hav­iour, in a shrewed spin on work of sci­ence fic­tion author Isaac Asimov.

He gave the ex­am­ple of a robot be­ing told to sim­ply fetch cof­fee.

A ma­chine not con­strained by proper prin­ci­ples might de­cide that ac­com­plish­ing the task re­quired it to de­fend against be­ing shut down and re­move all ob­sta­cles from its path by what­ever means nec­es­sary.

Rus­sell coun­selled robot prin­ci­ples, in­clud­ing al­tru­ism, hu­mil­ity, and making a pri­or­ity of hu­man val­ues.

“You are prob­a­bly bet­ter off with a ma­chine that is like this,” Rus­sell said.

“It is a first step in hu­man com­pat­i­bil­ity with AI.”

As ma­chines get smarter, so do we.

Newspapers in English

Newspapers from Malaysia

© PressReader. All rights reserved.