How Will GenAI Change the Consulting Industry?

Stories on the threat that generative AI poses to knowledge workers often focus what is likely to be known as “structured creativity”: duties like copywriting or laptop programming the place outputs are clearly specified by issues like artistic briefs or system necessities. One group that typically doesn’t think about itself at excessive danger are the analysts and consultants who write the stories themselves. In any case, their work requires insights and judgements which are past the present capabilities of genAI, which in the end does nothing greater than play again the coaching information it is fed.

Living proof: This widely reported BCG study discovered that genAI instruments considerably improved efficiency of consultants at product planning, with the best enhancements amongst below-average performers. A second discovering of the identical examine was much less broadly famous: AI depressed efficiency at producing insights from interviews and uncooked information. This dovetails properly with the concept that AI can assist with structured duties akin to marketing campaign administration however isn’t (but) able to growing new insights. One motive that AI struggles with insights is its well-known tendency to hallucinate, which might translate into false conclusions in regards to the that means of enter information. (See, for instance this BBC study that discovered 47% of AI chatbot solutions contained vital errors about information sources or accuracy.)*

So plainly the survey analyzers are certainly protected, not less than in the interim. However whereas insights are an necessary a part of the worth supplied by analysts and consultants, they’re only a small fraction of their work. In my expertise, the primary causes that corporations rent consultants are to entry their abilities at organizing a venture and their information of distributors to think about hiring.** These are issues that genAI does extraordinarily properly: ChatGPT will gladly present a believable venture plan for practically any job and can simply as merrily reply your questions on which distributors to think about, offering detailed comparisons on related options. It is true that these comparisons will usually be inaccurate (see the BBC examine or do your personal check in a subject you understand properly). However, let’s face it, consultants additionally make errors about product capabilities, which change a lot quicker than even probably the most diligent trade knowledgeable can sustain. So, whether or not a vendor record comes from genAI or a human guide, it nonetheless must be validated with present data earlier than you base any choices on it.

Much more to the purpose: simply because it’s all the time been onerous for purchasers to inform actually educated consultants from pretenders, it’s now equally onerous to inform correct genAI outputs from hallucinations. This is applicable to frameworks, vendor lists, and even these insights the place people are nonetheless clearly superior. GenAI is famously able to offering reasonable-sounding arguments primarily based on false premises, citing false proof, and resulting in mistaken conclusions. A real optimist would argue these can result in fruitful discussions even once they begin with an error. A extra sensible expectation is that some giant fraction of individuals will settle for them as delivered. 

That is very true in terms of feedback on trade traits and futures, the place there is no such thing as a clearly proper reply. A big a part of the trade dialog at this time, particularly in media like LinkedIn, is precisely that form of dialogue. AI-generated feedback are more and more frequent and, fact be instructed, may be hard to tell from the human-generated ones. Positive, that is a tricky scenario for trade members attempting to get good data. However avoid wasting sympathy for the poor human consultants who’re attempting to be heard – and acknowledged as respectable specialists – amid all of the noise. 

Certainly, the flood of content material is arguably the most important change in advertising and marketing in recent times. Not so way back, publishing good content material regularly was an efficient, low-budget path to success. The mechanism was easy: good content material was picked up through social media sharing and, if it grew to become standard sufficient, registered in search engine outcomes. At present, not solely is there an excessive amount of content material for anybody who’s not already well-known to realize a major share of consideration, however AI search summaries will scoop up good data with out main readers again to the supply. The choice path of publishing on high-traffic information websites is equally blocked since AI serps stifle site visitors to these websites as properly. In fact, a handful of influencers who’re expert at attracting consideration can nonetheless change into Web-famous.  However what good points consideration isn’t essentially what’s of true worth.***

I’ll skip the standard moaning about dangerous data driving out the great, for the reason that penalties are throughout us. As a substitute, I’ll posit that the scenario is untenable as a result of folks merely won’t be keen to devour such big volumes of content material indefinitely. There may be already a backlash amongst individuals who need to cut back the time they spend attempting to maintain up. (In truth, time spent on social media peaked in 2022.)  The plain resolution is to let AI view media for you and summarize what’s price seeing. Some folks might discover that acceptable however I’d wish to suppose that the majority will acknowledge that AI can’t be trusted to make the suitable selections – particularly if the AI is run by companies with pursuits of their very own. The choice would appear to be counting on human curators – keep in mind “editors?” — to evaluate the accessible supplies and choose these price consuming. In fact, deciding which curators to belief will itself require customers to do some work. This path additionally requires an alternative choice to the failing economics of conventional, ad-supported media, which is able to most likely be consumer-paid subscriptions that compensate each the content material creators and the curators. A center floor is likely to be AI curators which are funded by customers slightly than self-interested company suppliers.

I gained’t fake to know strategy will dominate. Maybe totally different teams will undertake totally different approaches, with entry to correct data changing into a luxurious good like ad-free TV. For now, folks can already be selective of their content material consumption by avoiding the undifferentiated (or, extra exactly, platform-selected) streams in information and social media feeds. The dilemma is sharper for specialists counting on content material creation to construct their enterprise: do they attempt to sustain with content material bots that may simply outproduce them, recognizing that even constantly good, distinctive, and insightful content material gained’t essentially be seen? Or do they discover different methods to succeed in potential purchasers, maybe by reside occasions or much less mediated platforms like reddit and medium?

Thoughts you, that is occurring in a world the place AI can legitimately exchange among the experience consultants are used to promoting, akin to methodologies, technical abilities, and vendor information. This results in smaller engagements the place purchasers are shopping for solely the AI oversight and inventive insights that people alone can present. So does the whole marketplace for consulting companies shrink at precisely the identical time that promoting these companies turns into tougher? Or, extra optimistically, does decrease value allow consultants to increase the market by promoting to companies that couldn’t beforehand afford their companies? (See, for instance, this new WPP offering of AI-enabled “self-service” advert marketing campaign improvement.) Whereas AI-powered companies don’t themselves create extra human jobs, they do presumably supply a possibility to promote insight-based tasks that do require paying people.

I am going to say it once more: the long run is unknown. However I do consider that analysts who’re calmly assessing the menace that AI poses to different folks’s jobs may give extra pressing thought to the menace AI poses to their very own. 

* I’ll ignore the excellence between AI augmenting staff and AI changing them: When you make folks twice as productive, then half the employees are not wanted. It’s true that there’s some added work to examine AI outcomes, however, regardless of complaints about ‘workslop’, we will most likely assume that there’s a considerable web labor financial savings. Definitely most AI users say they’re satisfied with the advantages regardless of the additional work. 

** Caveat: my expertise is biased since most of my consulting was round vendor choice. 

*** My entry within the Understatement of the 12 months competitors.