As search engines like Google, Bing, Face­book and Pin­ter­est increas­ing­ly use arti­fi­cial intel­li­gence (AI) to sur­face the most rel­e­vant con­tent and ensure adver­tise­ments don’t appear next to objec­tion­able con­tent, it’s clear AI still has a lot to learn.

But even as AI gets bet­ter, what does more AI-based con­tent sur­veil­lance mean for brands and mar­keters? And will it real­ly sur­face the best con­tent and resolve adver­tis­ers’ recent con­cerns about ad place­ments along­side objec­tion­able con­tent?

For one thing, experts say reliance on AI may pos­si­bly cre­ate even more algo­rith­mic hur­dles for search mar­keters.

How­ev­er, Yulia Khans­vyaro­va, head of dig­i­tal mar­ket­ing at SEM­rush, said whether the sit­u­a­tion gets bet­ter or worse for a giv­en brand depends on pre-exist­ing atti­tudes toward that brand.

“There is noth­ing to fear if most men­tions of the brand are pos­i­tive,” she said. “What can be said is that with the increase of AI usage by Google, it will be much hard­er to change users’ gen­er­al atti­tude towards the brand – no mat­ter how pos­i­tive or neg­a­tive it is.”

At the same time, she not­ed we still don’t know exact­ly how Google eval­u­ates con­tent qual­i­ty.

“We were told that we have to cre­ate user-friend­ly con­tent that gets shared and that behav­ioral fac­tors are play­ing a large role in con­tent rank­ing. What changes will come along with [AI]? It depends on the area where AI will be used and the scope of its usage,” Khans­vyaro­va said. “Will behav­ioral pat­terns be con­sid­ered as input vari­ables for machine learn­ing algo­rithms? Or will [AI] be used main­ly for nat­ur­al lan­guage pro­cess­ing tasks? For exam­ple, today, one of the most well-known cas­es of AI imple­men­ta­tion by Google is [machine learn­ing tech­nol­o­gy] RankBrain and RankBrain is used to find seman­ti­cal­ly relat­ed words and is based on algo­rithms sim­i­lar to word2vec.”

And accord­ing to Fly­nn Zaiger, CEO of dig­i­tal mar­ket­ing agency Online Opti­mism, what makes AI dif­fer­ent from pre­vi­ous algo­rithm changes is that it has the abil­i­ty to change far more quick­ly.

“Machine learn­ing should put the changes at the approval and behest of code, rather than search qual­i­ty teams,” Zaiger said. “There­fore, dig­i­tal mar­keters that spe­cial­ize in search will have to pay far more atten­tion to their port­fo­lio of web­sites to notice updates day to day.”

And then there are the pol­i­tics of search engines accept­ing third par­ty con­tent and plac­ing brand con­tent next to it with­out accept­ing any respon­si­bil­i­ty because they didn’t cre­ate it. One search indus­try insid­er used the exam­ple of a video that talks about Hitler being a nice guy, as well as about the hand cream he used, which could end up with pre-roll from John­son & John­son. That isn’t good for the brand, but Google could arguably claim rel­e­van­cy because John­son & John­son makes hand cream.

In fact, one les­son from the YouTube ad boy­cott could be programmatic/AI still present dan­gers because, as the New York Times not­ed, AI still doesn’t under­stand con­text and what dis­tin­guish­es good and bad con­tent, as we have seen in recent weeks.

Zaiger, too, said AI will con­tin­ue to have issues until it gets per­fect­ed, which may take years – if it ever hap­pens at all.

“There­fore, mon­i­tor­ing and report­ing of these sit­u­a­tions to keep your ads where you want them will be nec­es­sary for a while,” he said. “That being said, since AI can cause wide­spread changes sys­tem-wide, it will offer agen­cies who man­age clients in a vari­ety of indus­tries an advan­tage since they maybe notice the issue first and can adjust faster.”

Sim­i­lar­ly, John Caioz­zo, senior SEO ana­lyst at SEO agency SEO Inc., said increased use of AI means it’s even more crit­i­cal for brands to keep an eye out for neg­a­tive men­tions and con­tent.

“Google has always seemed to rank and pre­fer neg­a­tive con­tent and news around a brand. One of the rea­sons for this is the viral nature of the con­tent and the engage­ment it gen­er­ates mak­ing the result unique and rel­e­vant to the brand,” he said. “There is no guar­an­tee that AI will deliv­er the right con­tent to the right per­son at the time. One exam­ple of this right now is Google’s fea­tured snip­pet, [which] often returns ques­tion­able results from ques­tion­able sources, like [one] exam­ple where appar­ent­ly, Oba­ma was plan­ning a com­mu­nist coup.”

Fur­ther, Caioz­zo said in such cas­es where fea­tured snip­pets have returned bad results, Google has been quick to take action and cor­rect or remove the fea­tured snip­pet, but it is impos­si­ble for Google to police them all. In such cas­es, all brands can do is noti­fy Google and hope it is fixed in a time­ly man­ner.

“Hope­ful­ly in the future, Google will be able to refine its algo­rithms and AI to serve more true and reli­able results but this is not guar­an­teed and it will like­ly still be pos­si­ble to pro­lif­er­ate false infor­ma­tion about a brand or per­son,” he added.

How­ev­er, Khans­vyaro­va point­ed out Google uses dif­fer­ent approach­es for rank­ing organ­ic and paid con­tent.

“As for paid ads and ad rel­e­vance, I’m sure the sit­u­a­tion will change for the bet­ter and users will get the right con­tent at the right time and in the right place because paid con­tent, com­pared to organ­ic, has a real­ly sig­nif­i­cant advan­tage in terms of [AI] and machine learn­ing. The advan­tage is a clear­ly defined goal and action,” she added.

That’s because AI needs clear def­i­n­i­tions and paid con­tent can deliv­er. And if the algo­rithms used to find audi­ences that are sim­i­lar to a brand’s cus­tomers don’t work smooth­ly, adver­tis­ers will lose mon­ey, which will lead to ad bud­gets being decreased and a reduc­tion in Google’s income.

“Google can’t afford this, so I guess per­son­al­iza­tion and rel­e­vant con­tent deliv­ery will increase,” Khans­vyaro­va said.

Zaiger, too, is opti­mistic algo­rithms will become bet­ter judges of seman­tics and place con­tent more appro­pri­ate­ly over the long term.

“At that point, though, it’s a mat­ter of whether objec­tion­able con­tent can adjust as fast as the AI,” he said. “[Con­sid­er­ing] that it’s been years since I’ve seen an adult site slip through Google’s Safe­Search, I doubt they will.”