McKinsey report: State of AI in 2017 [pdf] | Hacker News Reader

McKinsey report: State of AI in 2017 [pdf]

mckinsey.com

82 pointsmatco1115 comments

The last time I talked to some "experts" at McKinsey about data science they said the leaders in the space were heavily investing in IBM's Watson and suggested we do the same (this was right before everything started hitting the fan with Watson). They don't have much credibility in this space, despite trying very hard to sound credible.

There are some smart people there but their "advice" rarely holds up to scrutiny when actual experts start poking at it--hence why they focus so heavily on vague McKinsey 'indexes' and 'benchmarks' that are very smoke and mirrors.

In the example above some McKinsey "experts" were presenting such BS to a room of people quite knowledgeable about this stuff and after the first few questions it was clear they really knew almost nothing about data science, ML or AI. Indeed even at that time telling a room full of actual experts that the future was Watson would quite literally get you laughed out of the room.

I think AI is way over-hyped. There are quite a few online tech publications right now in which more than half of all articles are AI-related; usually promoting some gimmicky chat-bot startup or some service that reads your emails to automate stuff (how can that possibly go wrong?), or some AI advertising service (as if advertising isn't targeted enough)... It all sounds useless to me.

The vision of the founders of a lot of these companies don't seem to extend beyond their own nose - They can't really see anything past that gigantic mountain of VC money.

It seems that for some reason AI has become the last avenue for growth and all VCs are just rushing to pour all their money into it.

Researchers with PhDs have been working on more or less the AI technologies for more than 30 years; it's been their life's work; throwing money all over the place isn't going to make them work any faster - It's just going to attract unqualified impostors into the industry.

As a founder, it's much easier to come up with a strategy to pocket a good chunk of that VC money for yourself than to actually solve the problem.

Spot on. It is amusing to watch though :) I think AI investors may be perpetuating the hype as it makes it harder for companies to not buy an AI product or two.. recipe for sharp correction down the road.

Is McKinsey really a good source for predictions? Do they have any in house expertise on machine learning (I doubt it)? And do they even have a track record of predictions for strategic, technological trends?

Well, this report isn't really focused on making predictions. It's more looking at adoption and investments and McKinsey does have a pretty good understanding of where enterprises are investing time and money for strategic initiatives.

As far as expertise, I suppose it depends on what you mean. Active AI practitioners. Probably not except to the degree that they pay outside experts for input. But they certainly have people who follow technology fields closely and have technology background. e.g. one of the authors of this report is a CS PhD, former CIO, and I've seen him speak at CIO conferences on a couple of occasions.

> McKinsey does have a pretty good understanding of where enterprises are investing time and money for strategic initiatives.

Isn't this just looking at investments in different areas and compiling a table?

Well, yes. It's just pulling data together from a bunch of different sources and interpreting it, talking to people, putting together case studies, presenting it in a fairly concise format, etc. Certainly other firms are capable of doing something similar but it's not as if this sort of information is readily available from a single source to just turn into a table.

What exactly are the interpretations beyond the obvious here?

For researcher and practitioners in the field? Probably not a lot. Although researchers may not have a lot of understanding about how the tech is actually being used. For the execs who are McKinsey clients or potential clients, it's probably a useful overview.

About to join their analytics team soon. Their technical capability is high and growing. Their involvement with "AI" is not that huge, but they'll have one of the best outlooks of how average companies are starting to throw machine learning/optimization at their operations.

They do have some incredibly talented people working for them, and they get to see a great many different organisations of every size, industry, and nationality, or type.

You can probably better find the technical state-of-the-art on arxiv github. But I can't think of many people or organisations better placed to know the impact of AI on business. The technology is developed at universities, and at the likes of Google and Facebook, who are focused on in-house applications only. Maybe IBM, but the last person to read an IBM white paper was later seen walking South from McMurdoch station and is feared(and/or hoped) dead.

Personally, I still prefer the Kinsey report. But this certainly isn't terrible.

Of course not.

"Only 20% of AI-aware firms are adopters." This number is much lower than I expected considering how routinely advances and applications of AI are reported in the tech community. There are so many opportunities for entrepreneurs on the horizon.

Someone should add a [pdf] to the title

The post title and linked PDF don't sound like the same publication.

The PDF title is "AI: The Next Digital Frontier? Discussion Paper June 2017"

Reply on news.ycombinator.com