Thought Leader In­ter­view:

Ajay Agrawal

Rotman Management Magazine - - FRONT PAGE - By Karen Chris­tensen

by Karen Chris­tensen

De­scribe what hap­pens at the Creative De­struc­tion Lab (CDL).

The CDL is a seed-stage pro­gram for mas­sively scal­able sci­ence­based com­pa­nies. Some start-ups come from the Univer­sity of Toronto com­mu­nity, but we now also re­ceive ap­pli­ca­tions from Europe, the U.S. (in­clud­ing Sil­i­con Val­ley), Is­rael and Asia.

We launched the pro­gram in Septem­ber 2012, and each au­tumn since, we’ve ad­mit­ted a new co­hort of start-ups into the pro­gram. Most com­pa­nies that we ad­mit have de­vel­oped a work­ing pro­to­type or proof of con­cept. The most com­mon type of founder is a re­cently grad­u­ated PHD in En­gi­neer­ing or Com­puter Sci­ence who has spent sev­eral years work­ing on a prob­lem and has in­vented some­thing at the fron­tier of their field.

The pro­gram does not guar­an­tee fi­nanc­ing, but the ma­jor­ity of com­pa­nies that suc­ceed raise cap­i­tal from the CDL’S Fel­lows and As­so­ciates — a care­fully- se­lected group of in­di­vid­u­als who them­selves are se­rial en­trepreneurs and early-stage in­vestors. Through­out the year, our MBA stu­dents work with the start-up founders as part of a sec­ond-year elec­tive course, help­ing them de­velop fi­nan­cial mod­els, eval­u­ate po­ten­tial mar­kets, and fine­tune their scal­ing strate­gies.

To date, more than 100 start-ups have come through the Lab. When we launched, we set a goal of gen­er­at­ing $50 mil­lion in eq­uity value cre­ated in terms of the ag­gre­gate value cre­ated by com­pa­nies that went through the Lab. When we fin­ished our fifth year in June 2017, we had ex­ceeded $1.4 bil­lion in eq­uity value cre­ated.

What ex­actly does the Lab pro­vide to en­trepreneurs?

Start-up founders ben­e­fit from a struc­tured, ob­jec­tives-ori­ented process that in­creases their prob­a­bil­ity of suc­cess. The process is or­ches­trated by the CDL team, while CDL Fel­lows and As­so­ciates gen­er­ate the ob­jec­tives. Ob­jec­tive-set­ting is a cor­ner­stone of the process. Ev­ery eight weeks the Fel­lows and As­so­ciates set three ob­jec­tives for the start-ups to achieve, at the ex­clu­sion of ev­ery­thing else. In other words, they de­fine clear goals for an eight week ‘sprint’. Ob­jec­tives can be busi­ness, techn­nol­ogy or Hr-ori­ented. Our Fel­lows and As­so­ciates—all vol­un­teers—are crit­i­cal to the CDL’S suc­cess.

Tell us more about the CDL Fel­lows and As­so­ciates.

We have de­signed a mar­ket­place — a com­mu­nity that func­tions un­der a set of rules and norms — that fa­cil­i­tates ef­fi­cient trans­ac­tions be­tween first-time founders and ex­pe­ri­enced en­trepreneurs, many of whom are also in­vestors. Of­ten, the two sides don’t know each other un­til the rookie founder seeks out the ex­pe­ri­enced en­tre­pre­neur/in­vestor when rais­ing cap­i­tal. Know­ing very lit­tle about the en­tre­pre­neur, the in­vestor usu­ally says no,

but oc­ca­sion­ally says yes, at which point they are very com­mit­ted. By the time we hit the end of the aca­demic year, the Fel­lows and As­so­ciates have met with the ven­tures many times, and they’ve got­ten to know each other. Fur­ther­more, the en­trepreneurs have demon­strated their abil­ity (or in­abil­ity) to de­liver against an ag­gres­sive set of ob­jec­tives through sev­eral cy­cles. We don’t re­quire Fel­lows and As­so­ciates to in­vest, but they can — and many do.

When you’re found­ing a com­pany, you’re faced with a thou­sand dif­fer­ent things you could be work­ing on. The ques­tion is, what should you fo­cus on? Peo­ple who have done this be­fore are able to triage those thou­sand things and pri­or­i­tize the two or three most im­por­tant things to fo­cus on right now to in­crease value and de-risk the ven­ture as quickly as pos­si­ble. Ev­ery eight weeks, this group meets with the founders and sets ob­jec­tives; then the founders carry on build­ing their ven­tures.

The CDL does not charge fees or take eq­uity. The cur­rency for par­tic­i­pa­tion is per­for­mance. There were seven Fel­lows in our first year, so the bi-monthly ses­sions were named af­ter that Group of Seven Fel­lows, or ‘G7’. At the end of each G7 ses­sion, we ask the Fel­lows and As­so­ciates to raise their hand for any com­pa­nies for which they are will­ing to com­mit their most pre­cious re­source: time. We ask them to com­mit to meet­ing with the com­pany for an hour ev­ery other week un­til the next ses­sion.

Any com­pa­nies that don’t in­spire at least one raised hand are dropped from fu­ture meet­ings — although they are still part of the CDL fam­ily and at­tend other events. The rule is that at least one com­pany must be dropped from the G7 ses­sions at each meet­ing. In the rare case that hands go up for all com­pa­nies, we raise the price in terms of the amount of time re­quired to com­mit. As we pro­ceed through­out the nine-month pro­gram, this al­lows us to fo­cus more and more re­sources on ven­tures that are show­ing the most progress.

Three years ago, CDL made a huge bet on ar­ti­fi­cial in­tel­li­gence (AI) and ma­chine learn­ing. What prompted that?

In our first year of oper­a­tion, one of the start-ups that came to us was Che­ma­tria, now called Atom­wise. Its founder, Abe Heifets — a U of T PHD in Com­puter Sci­ence and Bi­ol­ogy — was ap­ply­ing a new AI tech­nique to drug dis­cov­ery. What Abe was do­ing rep­re­sented not just a marginal im­prove­ment, but a po­ten­tially trans­for­ma­tive change to the way drugs are dis­cov­ered — which rep­re­sents a multi­bil­lion-dol­lar prob­lem for the phar­ma­ceu­ti­cal in­dus­try.

While we were work­ing with Abe, a team of grad­u­ate stu­dents from U of T Com­puter Sci­ence won a high-pro­file com­pe­ti­tion at Stan­ford called Ima­genet. It’s ba­si­cally an im­age-

We saw mount­ing ev­i­dence that AI was a gen­eral-pur­pose tech­nol­ogy that can be ap­plied to a wide range of prob­lems.

recog­ni­tion com­pe­ti­tion, whereby a com­puter is given a bunch of pic­tures and has to iden­tify the im­age, whether it’s a ball, a horse or a wheel­bar­row. This team from Toronto par­tic­i­pated, and not only did they win — us­ing a ma­chine learn­ing tech­nique called deep learn­ing, largely de­vel­oped at U of T — but they won by such a mar­gin that the fol­low­ing year, all of the fi­nal­ist teams were us­ing their tech­nique.

Those are just two ex­am­ples of events that in­spired us to bet on ma­chine in­tel­li­gence. Over­all, we saw mount­ing ev­i­dence that AI was a gen­eral-pur­pose tech­nol­ogy that could be ap­plied to a wide range of prob­lems across a vast ar­ray of in­dus­tries, and that’s what prompted us to ded­i­cate a new stream of the Lab’s ac­tiv­ity to AI.

At first, you faced re­sis­tance; why?

Peo­ple said we were be­ing too nar­row, that there weren’t enough star­tups to fill an AI stream and that there wasn’t enough in­ter­est from in­vestors. At the same time, we had be­liev­ers. One such be­liever who her­self had writ­ten a highly in­flu­en­tial blog post de­scrib­ing the ‘land­scape’ of com­pa­nies emerg­ing in the ma­chine learn­ing world was Shivon Zilis — a Cana­dian based in San Fran­cisco and a part­ner at the ven­ture in­vest­ing firm Bloomberg Beta, where she led the firm’s in­vest­ments in ma­chine in­tel­li­gence. I in­vited her to the Rot­man School to present her in­sight­ful anal­y­sis to our MBA stu­dents, and the CDL team — quickly re­al­iz­ing she is a star — re­cruited her to join forces on our AI ini­tia­tives. ( Elon Musk sub­se­quently saw the same po­ten­tial and re­cruited her to help him build his em­pire.)

So, we moved for­ward with the new stream, but to ad­dress these con­cerns, in 2015 we also launched an an­nual con­fer­ence — with Shivon as co-chair — called Ma­chine Learn­ing and the Mar­ket for In­tel­li­gence. The goal was to ed­u­cate the Cana­dian busi­ness com­mu­nity about the im­por­tance of this emerg­ing field. Lead­ers in the field — from or­ga­ni­za­tions like Google, Uber, Ap­ple, Stan­ford, Carnegie Mel­lon and MIT — came to Rot­man to dis­cuss and de­bate how AI is and will im­pact a va­ri­ety of fields, from life sciences to man­u­fac­tur­ing to re­tail. We held our third an­nual con­fer­ence in Oc­to­ber 2017.

Talk a bit about the CDL’S re­sults to date.

The launch of our AI stream trans­formed the Lab from a Cana­dian en­ter­prise into a global one. In our first year, our start-ups were all from On­tario, but they now come from around the world. Sim­i­larly, in our first year, our Fel­lows were all from Canada, and that, too, changed when we launched the AI stream. Our ML7 (Ma­chine Learn­ing Seven) in­cludes Wil­liam

Tun­stall-pe­doe, who flies in from Cam­bridge, Eng­land, ev­ery eight weeks. He has a PHD in Ma­chine Learn­ing and founded Evi, which was ac­quired by Amazon in 2012. Evi’s tech­nol­ogy pow­ers the AI en­gine in Amazon’s Alexa, which, to my knowl­edge, is still the top-sell­ing con­sumer AI hard­ware prod­uct in the world.

The ML7 also in­cludes Bar­ney Pell, who flies in ev­ery eight weeks from San Fran­cisco. Bar­ney also has a PHD in Ma­chine Learn­ing and led an 85-per­son team at NASA that flew the first AI into deep space. He then built an AI com­pany called Pow­er­set that was ac­quired by Mi­crosoft, and now he’s the co­founder of Moon Ex­press, which is es­sen­tially build­ing a Fed­eral Ex­press- type ser­vice to the moon, be­cause Bar­ney be­lieves the moon is go­ing to be an im­por­tant gate­way for com­mer­cial space travel.

So far, the re­sults have sur­passed our ex­pec­ta­tions. Back in 2012, we ac­cepted 25 com­pa­nies into our gen­eral high-tech stream. Last year, we dou­bled that by adding the sec­ond co­hort fo­cused on AI, so we had 50 start-ups. This year, we dou­bled our in­take again by ac­cept­ing 100 Ai-fo­cused start-ups and adding a new stream: The world’s first pro­gram fo­cused on launch­ing star­tups pred­i­cated on quan­tum ma­chine learn­ing (QML). To our knowl­edge, the CDL is home to the great­est con­cen­tra­tion of Ai-based com­pa­nies of any pro­gram on Earth.

Among CDL’S ‘grad­u­at­ing’ com­pa­nies to date, which best per­son­i­fies your vi­sion?

We are proud of all of them, and dif­fer­ent com­pa­nies re­flect dif­fer­ent as­pects of our vi­sion. For ex­am­ple, Atom­wise per­son­i­fies our fo­cus on the ap­pli­ca­tion of sci­ence that can have a trans­for­ma­tive ef­fect on so­ci­ety. As in­di­cated ear­lier, it brought in a very early ap­pli­ca­tion of a new branch of Com­puter Sci­ence (deep learn­ing) and ap­plied it to a com­mer­cial fo­cus (drug dis­cov­ery).

Thalmic Labs cap­tures the scale and am­bi­tion of our mis­sion. They raised their seed fi­nanc­ing largely from our G7 Fel­lows. About a year ago, they raised $160 mil­lion in Se­ries B fi­nanc­ing (US$120 mil­lion), which was one of the largest Se­ries B fi­nanc­ings in Cana­dian his­tory.

UDIO, founded by Katya Ku­dashk­ina (Rot­man MBA ‘15), cap­tures the CDL’S en­tre­pre­neur­ial spirit: She im­mi­grated to Canada with­out a penny to her name and re­ally hus­tled to get into the top busi­ness school in the coun­try while study­ing English at night. When she grad­u­ated, she was re­cruited into a nice, se­cure job at the Cana­dian Pen­sion Plan In­vest­ment Board. She left that job to found a start-up, which she brought to the Lab, rais­ing a few hun­dred thou­sand dol­lars in in­vest­ment cap­i­tal.

Early on, UDIO was fo­cused on build­ing ro­botic bees for ar­ti­fi­cial pol­li­na­tion in the agri­cul­tural in­dus­try. But the G7 ad­vised Katya that it was go­ing to take too long to get to rev­enue with that busi­ness model, and that she needed to be closer to her cus­tomers. So, she packed up her life and moved to Cal­i­for­nia — es­sen­tially liv­ing on some­one’s couch so she could fo­cus on al­mond farm­ers in North­ern Cal­i­for­nia and learn their busi­ness from the ground up.

Ul­ti­mately, Katya ran out of cap­i­tal be­fore she could get to rev­enue and the com­pany folded; but she wouldn’t give up and launched another start-up. She learned a lot from her first com­pany and main­tained great re­la­tions with her in­vestors. I wouldn’t be sur­prised if they in­vest in her again, be­cause she is so driven, trust­wor­thy and will­ing to learn. She is a prime ex­am­ple of the per­sis­tence re­quired of en­trepreneurs.

As in­di­cated, CDL fea­tures col­lab­o­ra­tion with both cur­rent MBA stu­dents and highly suc­cess­ful en­trepreneurs. Can you give an ex­am­ple of a firm that ben­e­fited from both?

One ex­am­ple is Valid­ere. One of the co-founders just fin­ished his PHD at Har­vard, where he de­vel­oped a tech­nique called Op­ti­cal Liq­uid Finger­print­ing, which iden­ti­fies the prop­er­ties of a liq­uid. Nor­mally if you want to find out a liq­uid’s prop­er­ties, you take a sam­ple, send it to a lab, and wait for the re­sults. This startup de­vel­oped a process whereby they can es­sen­tially de­ter­mine the prop­er­ties of a liq­uid in real time. They came to the CDL want­ing to sell this ser­vice to the lux­ury per­fume in­dus­try, to help de­tect coun­ter­feit per­fumes.

Af­ter re­view­ing their busi­ness, the Fel­lows told the founders that they loved the tech­nol­ogy but hated the busi­ness idea. So, the G7 turned to our MBA stu­dents and asked them to do a mar­ket anal­y­sis to find out where this tech­nol­ogy would have the most value. The MBAS re­turned with a rec­om­men­da­tion to move from one of the sex­i­est of all in­dus­tries — lux­ury per­fume — to one of the least sexy: oil and gas. And that is what they ended up do­ing.

One of our Fel­lows is Dr. Chen Fong, for­mer head of Ra­di­ol­ogy at the Univer­sity of Cal­gary and ac­tive in­vestor in med­i­cal tech­nolo­gies as well as the en­ergy in­dus­try. Af­ter learn­ing about Valid­ere’s tech­nol­ogy and the rec­om­men­da­tion to fo­cus on oil and gas, Chen flew the founders out to Cal­gary and drove them (him­self!) around the city to meet with a num­ber of oil and gas ex­ec­u­tives to so­licit feed­back on their prod­uct.

Soon, the busi­ness was trans­formed: They at­tracted some sig­nif­i­cant cus­tomers and went from be­ing un­able to raise cap­i­tal to be­ing over-sub­scribed, with more in­vestor in­ter­est than

The Creative De­struc­tion Lab is home to the great­est con­cen­tra­tion of Ai-based com­pa­nies of any pro­gram on Earth.

they could ac­com­mo­date. Our MBA stu­dents learned a tremen­dous amount — a very dif­fer­ent but com­ple­men­tary type of ed­u­ca­tion than they get from read­ing about his­tor­i­cal case stud­ies. This ex­am­ple epit­o­mizes CDL’S vi­sion: A sci­ence-based in­no­va­tion that will en­hance so­ci­ety; cre­ated by ap­pre­cia­tive, per­sis­tent, and coach­able founders; na­tional con­nec­tions, in­sight­ful ad­vice and men­tor­ing from our G7 Fel­lows; and an in­cred­i­ble hands-on learn­ing ex­pe­ri­ence for our MBA stu­dents.

The Lab is one of the most pop­u­lar sec­ond-year MBA cour­ses at the Rot­man School. Why does it res­onate so much with stu­dents?

For two rea­sons: First, it com­bines the tra­di­tional mode of learn­ing from lec­tures with learn­ing-by-do­ing; and sec­ond, it links aca­demic work with a sense of own­er­ship.

The tra­di­tional ap­proach to learn­ing at CDL is led by our Chief Econ­o­mist, Pro­fes­sor Joshua Gans, who de­vel­oped a struc­ture for teach­ing en­tre­pre­neur­ial strat­egy along with MIT’S Scott Stern. This pro­vides stu­dents with an aca­demic frame­work and con­text for what they’re go­ing to ex­pe­ri­ence next. Then comes the learn­ing-by-do­ing part. Nor­mally, busi­ness schools use Har­vard Busi­ness School cases to pro­vide ex­am­ples in the class­room. As in­di­cated, we re­place those with real com­pa­nies. Work­ing with founders, Fel­lows and As­so­ciates pro­vides stu­dents with an op­por­tu­nity to roll up their sleeves. In­stead of read­ing a 30-page case that comes with a fact set, they have to find the facts them­selves and fig­ure out — of the in­fi­nite in­for­ma­tion out there, which bits are the most valu­able for their needs? They ex­pe­ri­ence the messi­ness of the real world and the re­al­ity of hav­ing to make de­ci­sions with­out hav­ing full in­for­ma­tion.

The sec­ond piece is own­er­ship. When our stu­dents work with these start-ups, ev­ery de­ci­sion mat­ters, so they have a real sense of own­er­ship. It’s a pow­er­ful learn­ing ex­pe­ri­ence to feel own­er­ship over the re­sults be­cause the con­se­quences are so tan­gi­ble.

Uni­ver­si­ties rarely adopt pro­grams de­vel­oped else­where. What mo­ti­vated the Univer­sity of Bri­tish Columbia, New York Univer­sity, the Univer­sity of Cal­gary, Univer­sité de Mon­treal and Dal­housie Univer­sity to adopt the Creative De­struc­tion Lab pro­gram?

Ev­ery univer­sity has a pro­gram or course on en­trepreneur­ship and start-ups, but I think the CDL stands out due to its sig­nif­i­cant re­sults. The cal­i­bre of in­vestors from the busi­ness com­mu­nity who have ral­lied around the CDL is un­prece­dented. Nat­u­rally,

Our ap­proach is to get ahead, make the in­vest­ments, and at­tract all the el­e­ments of the ecosys­tem to Canada.

other uni­ver­si­ties would love for that to hap­pen at their own busi­ness schools.

When UBC in­di­cated in­ter­est in adopt­ing the pro­gram, the big ques­tion was, ‘Is this repli­ca­ble?’ But a very com­pe­tent team, un­der the di­rec­tion of Pro­fes­sor Paul Cub­bon, was able to re­pro­duce it. When CDL-WEST com­pleted its first year, the re­sults on all di­men­sions were im­pres­sive, and we had ev­i­dence that, yes, this pro­gram is repli­ca­ble. We have since launched CDL at the Univer­sity of Cal­gary, Dal­housie Univer­sity and Univer­sité de Mon­treal, and in Oc­to­ber, we an­nounced a part­ner­ship with New York Univer­sity’s Stern School of Busi­ness.

CDL Toronto’s com­pe­ti­tion is not Van­cou­ver, Cal­gary, Mon­treal, New York or At­lantic Canada: it’s Sil­i­con Val­ley. Each of the CDLS has at­tracted some of the top busi­ness peo­ple from its re­gion. Our chal­lenge now is to cross-pol­li­nate, so that the Mon­treal Fel­lows are con­nect­ing with com­pa­nies in the Toronto pro­gram and the Toronto Fel­lows are con­nect­ing with com­pa­nies at CDL At­lantic, and so on. One of the things that makes the Bay Area so ef­fec­tive is that ev­ery­thing moves so fast. If we can ac­cel­er­ate the ve­loc­ity of busi­ness devel­op­ment here, we will have raised Canada’s game as a whole.

You men­tioned ear­lier that CDL launched the world’s first pro­gram fo­cused on quan­tum ma­chine learn­ing (QML). What is your vi­sion for this ini­tia­tive?

It’s a bold one: By 2022, the QML Ini­tia­tive will have pro­duced more well-cap­i­tal­ized, rev­enue-gen­er­at­ing quan­tum ma­chine learn­ing-based soft­ware com­pa­nies than the rest of the world com­bined, with the ma­jor­ity based in Canada.

Why QML? First, we can lever­age the lead­er­ship that CDL cur­rently has in the com­mer­cial ap­pli­ca­tion of ma­chine learn­ing. Sec­ond, we can lever­age Canada’s lead­er­ship in quan­tum com­put­ing at places like the Perime­ter In­sti­tute and the In­sti­tute for Quan­tum Com­put­ing in Water­loo, Univer­sité de Sher­brooke in Que­bec, and D-wave in Van­cou­ver, among oth­ers. Third, we can lever­age the net­work of in­vestors, en­trepreneurs, sci­en­tists, and cor­po­ra­tions that have ral­lied around the CDL and our mis­sion of com­mer­cial­iz­ing sci­ence for the ben­e­fit of hu­mankind. [ Ed­i­tor’s Note: For de­tails on the QML pro­gram, visit cre­ativedestruc­tion­lab.com/lo­ca­tions/toronto/quan­tum.]

Clearly, the CDL is lead­ing the way in this arena.

I be­lieve so. Three years ago, it felt like we were mov­ing early on AI, but we re­al­ize now that — if we could turn back the clock — we ac­tu­ally should have started even sooner and moved faster. We were roughly a year ahead of ev­ery­one else, but now a num­ber of

pro­grams in other coun­tries are fo­cused on AI star­tups — so we’re run­ning fast just to keep our po­si­tion.

In terms of QML, so far we’re the only ones do­ing it, and that’s be­cause the tech­nol­ogy is so em­bry­onic. We might go for two or three years with­out a sig­nif­i­cant suc­cess, be­cause we might be too early. The point is, once there’s a hit, places like MIT, Stan­ford and Sil­i­con Val­ley will all dou­ble down in this field. Our ap­proach is to get ahead, make the in­vest­ments now, and at­tract all the el­e­ments of the ecosys­tem to Canada.

We want to do in Toronto with QML what Sil­i­con Val­ley did with semi­con­duc­tors in the 1960s. There’s noth­ing in­her­ently mag­i­cal about Sil­i­con Val­ley. The semi­con­duc­tor in­dus­try hap­pened to start there due to the pi­o­neer­ing ef­forts of a hand­ful of peo­ple, and once that com­mu­nity grew big enough, it be­came very hard for other re­gions to com­pete. Our view is, if we can seed it here and if the in­dus­try takes off five years from now, by that time, Canada will have such a crit­i­cal mass that it will be hard for the whole com­mu­nity to move some­where else. We’re try­ing to plant the seeds now.

Al­ready, three top Sil­i­con Val­ley ven­ture cap­i­tal­ists are suf­fi­ciently op­ti­mistic about this pro­gram that they of­fered to in­vest in ev­ery one of the com­pa­nies that gets into it — sight un­seen. Most of these com­pa­nies won’t make it — and they know that — but they want to be in­volved be­cause along the way, they will get an ed­u­ca­tion in QML, and there is some pos­i­tive prob­a­bil­ity that one or two of these com­pa­nies will fig­ure out a com­mer­cial ap­pli­ca­tion.

Glob­ally, what has been the re­cep­tion to AI?

Ear­lier this year, the Cana­dian gov­ern­ment made a se­ries of fi­nan­cial com­mit­ments to at­tempt to main­tain its po­si­tion as a leader in AI re­search. In July, China an­nounced a long-term AI plan that dwarfs Canada’s in­vest­ment and spec­i­fies a time­line through 2030 dur­ing which China aims to be­come the world leader in AI. Over the Labour Day week­end, Rus­sian Pres­i­dent Vladimir Putin fore­shad­owed sig­nif­i­cant in­vest­ments when he stated: “AI is the fu­ture, not only for Rus­sia, but for all hu­mankind. Who­ever be­comes the leader in this sphere will be­come the ruler of the world.”

There has also been cau­tion: Tesla CEO Elon Musk has made pleas for gov­ern­ments to take AI safety se­ri­ously and to set up reg­u­la­tory bod­ies to man­age it. His Twit­ter re­sponse to Putin’s re­marks was: “It be­gins…,” which he fol­lowed with: “China, Rus­sia, soon all coun­tries w strong com­puter sci­ence. Com­pe­ti­tion for AI su­pe­ri­or­ity at na­tional level most likely cause of WW3 imo.” Musk is con­cerned that AI is de­vel­op­ing more rapidly than

With QML, we want to do in Toronto what Sil­i­con Val­ley did with semi­con­duc­tors in the 1960s.

we re­al­ize and that there is sig­nif­i­cant risk to hu­man civ­i­liza­tion. He feels it needs to be reg­u­lated, not un­like com­mu­ni­ca­tions, air traf­fic, fi­nan­cial ser­vices, health­care and aero­space.

Will AI change the way de­ci­sions are made in or­ga­ni­za­tions?

Yes. Ev­ery dis­rup­tive tech­nol­ogy low­ers the cost of some­thing, and in the case of AI, that some­thing is prediction. By prediction, I mean us­ing data that you have to gen­er­ate data that you don’t have. Eco­nomic the­ory tells us that as the cost of ma­chine prediction falls, we will use more of it. Prediction is an in­put to de­ci­sion-mak­ing un­der uncer­tainty. When faced with uncer­tainty, we need to pre­dict the like­li­hood of dif­fer­ent out­comes when we make a de­ci­sion. As ma­chine prediction be­comes cheaper, we’ll in­creas­ingly sub­sti­tute hu­man prediction for ma­chine prediction in the de­ci­sion-mak­ing process.

How­ever, prediction is not the only in­gre­di­ent for de­ci­sion­mak­ing. Judg­ment — the as­sign­ment of value or pay­offs to pos­si­ble out­comes — is also im­por­tant. Ma­chines do prediction, but only hu­mans have judg­ment. I an­tic­i­pate that or­ga­ni­za­tions will en­gage in much more de­ci­sion-mak­ing be­cause a key in­gre­di­ent is now much cheaper, and the value of hu­man judg­ment will in­crease, as we de­mand more of it. We can only spec­u­late on the as­pects of judg­ment that will be most valu­able, but things like eth­i­cal judg­ment, emo­tional in­tel­li­gence and artis­tic abil­ity are likely sus­pects.

Ajay Agrawal is the Peter Munk Pro­fes­sor of En­trepreneur­ship, Pro­fes­sor of Strate­gic Man­age­ment, and founder of the Creative De­struc­tion Lab at the Rot­man School of Man­age­ment. He is also co­founder of Next Canada, which in­cludes Next AI, a not-for-profit pro­gram to in­spire young en­trepreneurs and tech­nol­o­gists to ex­plore com­mer­cial op­por­tu­ni­ties that are a di­rect re­sult of ad­vances in AI. Along with Rot­man Pro­fes­sors Joshua Gans and Avi Gold­farb, he is co-au­thor of Prediction Ma­chines: The Sim­ple Eco­nom­ics

of Ar­ti­fi­cial In­tel­li­gence (forth­com­ing, Har­vard Busi­ness School Press, April 2018). For more on the book visit pre­dic­tion­ma­chines.ai

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.