Investing in artificial intelligence — is it the right time to buy?

If robots are tak­ing our jobs, they may also sup­port our retire­ment. Bank of Amer­i­ca Mer­rill Lynch esti­mates that 47 per cent of US jobs and 33 per cent of UK jobs have the poten­tial to be auto­mat­ed. But the tech­nol­o­gy that could fill those jobs – robot­ics and arti­fi­cial intel­li­gence – is both a threat to some work­ers and a poten­tial area of growth for investors, notably pen­sion funds, seek­ing to cap­i­talise on the new wave of indus­tri­al­i­sa­tion.

“It’s the next gen­er­a­tion,” says Philippe Cerf, Europe, Mid­dle East and Africa co-head of the tech­nol­o­gy, media and telecom­mu­ni­ca­tions group at invest­ment bank Cred­it Suisse. “Machine-learn­ing is applied to some extent across all tech these days.”

Pro­fes­sion­al investors are tak­ing a grow­ing inter­est in firms that are either exposed to or are based upon AI tech­nol­o­gy

Tak­ing an invest­ment posi­tion on the tech­nol­o­gy asso­ci­at­ed with a spe­cif­ic trend such as this, called the­mat­ic invest­ment, first requires the theme to be well defined. As arti­fi­cial intel­li­gence (AI) is an umbrel­la term, cov­er­ing a range of spe­cif­ic tech­nolo­gies, under­stand­ing what qual­i­fies is the start­ing point for any investor.

Defining artificial intelligence

Dun­can Ander­son, chief tech­nol­o­gy offi­cer for Wat­son Europe at IBM, says there are three aspects of tech­nol­o­gy that can fall into the cat­e­go­ry of cog­ni­tive com­put­ing, which IBM uses to avoid the sci­ence-fic­tion con­no­ta­tions of AI.

“The first aspect is under­stand­ing that typ­i­cal­ly includes nat­ur­al lan­guage recog­ni­tion, but equal­ly it could be image recog­ni­tion,” he says. “Try­ing to make sense of data com­ing into a sys­tem that a tra­di­tion­al com­put­er sys­tem could not under­stand.”

The sec­ond ele­ment is rea­son­ing, for exam­ple work­ing out the mean­ing of a ques­tion then form­ing a field of hypothe­ses about the answer. “That might also be work­ing out which of those you might go with,” he says.

“The third cat­e­go­ry would be machine-learn­ing, the abil­i­ty for it to get bet­ter and to learn from expe­ri­ence, not just by being pro­grammed.”

Exposure to AI technologies

Pro­fes­sion­al investors are tak­ing a grow­ing inter­est in firms that are either exposed to or are based upon AI tech­nol­o­gy, although most ana­lysts have dif­fer­ent but over­lap­ping uni­vers­es of com­pa­nies that are put in the AI buck­et. Bank of Amer­i­ca Mer­rill Lynch research indi­cates some $2 bil­lion was invest­ed in 322 pub­licly list­ed com­pa­nies exposed to AI-based tech­nolo­gies in 2014.

Ana­lyst firm Ven­ture Scan­ner, which itself uses pro­pri­etary machine-learn­ing algo­rithms to index, search and clus­ter sim­i­lar com­pa­nies, tracks a small­er uni­verse of 910 specif­i­cal­ly AI-focused firms across 63 coun­tries which have raised a total of $3.68 bil­lion in fund­ing from a range of sources.

Sar­b­jit Nahal, equi­ty strate­gist at Bank of Amer­i­ca Mer­rill Lynch, says: “This mar­ket is going to take off over the next five years. To date we have seen a basic lev­el, but I think that we are real­ly going to see an uptake of machine-learn­ing, nat­ur­al lan­guage usage and image recog­ni­tion with the dif­fer­ent fac­tors start­ing to come togeth­er. An inte­grat­ed pack­age in terms of machine-learn­ing and the con­cept of intel­li­gent machines will be going towards 2020.”

Pos­si­ble invest­ment oppor­tu­ni­ties stretch from direct invest­ment in firms that are whol­ly or par­tial­ly exposed to AI, buy­ing into invest­ment vehi­cles which are tar­get­ing invest­ments, to crowd­fund­ing star­tups. In many cas­es ven­ture cap­i­tal (VC) firms are fund­ing these busi­ness­es from their ear­ly stages of devel­op­ment to the point of going pub­lic or being acquired by larg­er firms.

Caution recommended

Investors should be cau­tious, warns Mike Lynch, founder of Invoke Cap­i­tal Part­ners, a VC fund with £1 bil­lion in assets, which has invest­ed in sev­er­al firms that use advanced pat­tern-match­ing algo­rithms and machine-learn­ing.

“There are a rel­a­tive­ly small num­ber of peo­ple who can devel­op machine intel­li­gence, so when a start­up uses the phrase you have to be care­ful as to whether there real­ly is machine intel­li­gence involved,” he says. “The major­i­ty of things I am sent that claim to be con­duct­ed using arti­fi­cial intel­li­gence are done with humans.”

Dr Lynch, who was co-founder and for­mer chief exec­u­tive of Auton­o­my, which pro­vides intel­li­gent search tools based on prob­a­bil­i­ty ana­lyt­ics, notes that pub­licly list­ed com­pa­nies will typ­i­cal­ly have cus­tomers using sys­tems. This pro­vides investors with the oppor­tu­ni­ty to mea­sure suc­cess. For star­tups test­ing their sys­tems at demon­stra­tions is cru­cial.

“A lot of peo­ple are going to lose a lot of mon­ey on things that look great, but don’t work in the real world,” he says.

Mr Nahal also cau­tions investors to exam­ine the mar­ket care­ful­ly. “When we talk about the risks, I think there can be a ten­den­cy to over­es­ti­mate the impact of tech­nol­o­gy in the short term and under­es­ti­mate on a long-term basis,” he says. “Short­er term, it’s going to be more oppor­tu­ni­ty than risk, allow­ing mul­ti­ple sec­tors to deal with issues like wage infla­tion, pro­duc­tiv­i­ty, and qual­i­ty of prod­ucts and ser­vices.”

BUY INTELLIGENTLY

When machine-learn­ing com­pa­nies are suc­cess­ful, they tend to gain an ear­ly lead over oth­ers as the data they accu­mu­late allows them to then leap ahead, just as big play­ers like Google and Face­book have done pre­vi­ous­ly.

“The tech­nol­o­gy learns from expe­ri­ence – this is the rev­o­lu­tion,” says Dr Mike Lynch, founder of Invoke Cap­i­tal Part­ners. “Who­ev­er gets the data, their algo­rithms learn more and they get bet­ter. You get a very pow­er­ful feed­back loop.”

But beyond the larg­er firms that are research­ing and invest­ing in arti­fi­cial intel­li­gence or AI, oth­er oppor­tu­ni­ties might present them­selves to investors.

McK­in­sey Glob­al Insti­tute analy­sis has found that cer­tain sec­tors are like­ly to be affect­ed as AI auto­mates knowl­edge work. These are:

• Finance – automat­ing rou­tine tasks, such as answer­ing cus­tomer calls or dis­patch­ing assis­tance
• Edu­ca­tion – extend­ing capa­bil­i­ties of pro­fes­sion­als, for exam­ple doc­tors and lawyers, with machine-learn­ing sys­tems that can spot con­nec­tions humans would miss
• Health­care – expand­ing access to advanced IT tools and oth­er infor­ma­tion sys­tems through nat­ur­al lan­guage inter­faces
• Media and com­mu­ni­ca­tions – pro­vid­ing auto­mat­ic con­tent cre­ation and syn­the­sis
• Gov­ern­ment and social sec­tors – rais­ing pro­duc­tiv­i­ty and effi­cien­cy by increas­ing con­sis­ten­cy of tasks, such as search­ing and analysing infor­ma­tion.

The researchers at McK­in­sey also high­light infra­struc­ture and util­i­ties, trans­porta­tion and retail as sec­tors that may be impact­ed.

For investors look­ing for pro­fes­sion­al invest­ment sup­port in iden­ti­fy­ing oppor­tu­ni­ties, there are sev­er­al options. “There are a num­ber of funds that have either inte­grat­ed this as part of a larg­er focus on automa­tion, dis­rup­tive tech­nol­o­gy or inno­v­a­tive tech­nolo­gies,” says Sar­b­jit Nahal, equi­ty strate­gist at Bank of Amer­i­ca Mer­rill Lynch. “Or there are a cou­ple of ded­i­cat­ed funds that have been set up in this space, as well as an exchange-trad­ed fund in the Unit­ed States.”

Defining artificial intelligence

Exposure to AI technologies