Trust Sci­ence’s Ashif Mawji is hop­ing his al­go­rithms com­bined with ar­ti­fi­cial in­tel­li­gence can re­veal the an­swer

Alberta Venture - - The Briefing -

>>> Ashif Mawji, se­rial en­tre­pre­neur and key­note speaker at this year’s Fast Growth 50 con­fer­ence, is build­ing a dig­i­tal plat­form to de­ter­mine the trust­wor­thi­ness of peo­ple, places and busi­nesses. No, this is not like that sin­is­ter app, Peeple (which would see peo­ple rat­ing each other as they might a lo­cal restau­rant). Trust Sci­ence, he in­sists, is trust­wor­thy.

AL­BERTA VEN­TURE: How does it work? ASHIF MAWJI: We look at about 38 data points, ev­ery­thing from your so­cial me­dia ac­tiv­ity to court data to trans­ac­tional data, and the pro­pri­etary al­go­rithms draw in­fer­ences. Then there are pa­ram­e­ters: Let’s say I want to lend you $1,000. It’s go­ing to tell me how trust­wor­thy you are in terms of pay­ing that back. If I change the loan to $1,000,000, it will give me a dif­fer­ent num­ber. The same thing goes for re­cruit­ing or dat­ing or us­ing the shared econ­omy.

AV: So I can punch in some­one’s name and get feed­back? AM: Yes, and you can get a con­text score, which lets you say, “I want to hire Ashif to be my gen­eral coun­sel,” and in that case my trust­wor­thi­ness level will come lower be­cause I’ve never been a gen­eral coun­sel. I don’t even have a law de­gree. But if you say you want Ashif to be an en­tre­pre­neur in res­i­dence, well my trust score will come in higher.

AV: How does it work for places and busi­nesses? AM: Let’s say I want to move into a par­tic­u­lar neigh­bour­hood. The data points there will be the houses bought and sold, crime data and stuff like that. For busi­nesses, we’ll look at prod­uct re­calls, law­suits, on­line ac­tiv­ity and re­turn poli­cies. AV: Is the data out there good enough? AM: You’d be sur­prised what’s out there, and it’s all pub­lic. You just have to dig. A lot of the things we’re do­ing, some­one with good on­line skills could do man­u­ally, but it would take you two or three days to check one per­son out. You use our sys­tem and you get it in five sec­onds.

AV: You’re not ask­ing for feed­back or in­put from peo­ple? AM: We don’t be­lieve in feed­back or rat­ings be­cause we think that can be gamed. We think there’s enough data out there so we can ob­jec­tively cal­cu­late. Let’s say there’s a news­pa­per ar­ti­cle on me. You could have 100 peo­ple make com­ments. We read those com­ments but most im­por­tantly we look at who is mak­ing those com­ments and how trust­wor­thy they are in that field. If it’s a story on ar­ti­fi­cial in­tel­li­gence and a pro­fes­sor who teaches AI makes a com­ment, that’s go­ing to have a much higher cred­i­bil­ity than Joe Blow talk­ing about AI.

AV: How does the AI play into Trust Sci­ence? AM: You’re ba­si­cally train­ing the ma­chine. It’s look­ing at pat­terns and say­ing, “Th­ese pat­terns re­sult in this ac­tion.” So you throw a whole bunch of data at it and it ap­plies the in­tel­li­gence to say, “If those pat­terns equal X, then th­ese pat­terns equal Y.”

AV: Is Trust Sci­ence com­mer­cially avail­able? AM: We’re in a beta one: A few peo­ple have it. Beta two will be out around Fe­bru­ary. We have banks, re­cruit­ing firms and al­ter­na­tive lenders lined up. We have law-en­force­ment and shared-econ­omy part­ners. We’re get­ting our agree­ments in place and they’ll start beta two when it’s re­leased. Then we’ll be com­mer­cial around Q2 of next year. – Michael Gan­ley

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.