Enterprise hits and misses - the potential and limits of AI, and the cloud ambitions of Google
- Summary:
- In this edition: grasping AI's limits/possibilities through field lessons and use cases. Also: Google's push to become an enterprise cloud player gets a close look at Google Cloud Next. International Women's Day raises questions on women in tech progress. Your whiffs include bad PR promotions, an ill-fated BBC home interview, and the most jargon-infected blog post title of the year so far.
diginomica hit: The potential and limits of AI - use cases by Phil Wainewright
quotage: "That’s why this demonstration so perfectly encapsulates both the potential and the limitations of the current state-of-the-art in artificial intelligence. AI only works well when the parameters are well defined. Point it at a standardized cabinet with a limited set of products to recognize and it won’t take long to learn what it has to do because there aren’t that many “custom classifiers” that it has to deal with." - Salesforce captures the limits of AI in a Coca-Cola cooler
myPOV: In a pair of Salesforce use cases, Phil examines the limits of AI - and finds that the best AI approaches tend to be very specific. In the Coca-Cola piece, Phil shares a recent Salesforce Einstein demo, presented at Salesforce offices and streamed to the world. In this demo, Einstein uses its visual recognition engine, Einstein Vision, to recognize and count the variety and quantity of Coca-Cola in a cooler display, simple from an iPad or iPhone photo. The implications? No need for a Coca-Cola rep to physically visit and count. Einstein can also predict replenishment needs.
But as Phil points out, this example also shows the limits of "AI," with the current inability to count depth of rows in the storage cabinet, which itself is rigid in its flavor varieties: "The visual recognition Einstein is being asked to do in this demonstration is hugely constrained." Phil's next use case highlights Einstein's Watson integration, What Kellogg learned from Bear Naked's AI-powered e-commerce portal.
This is another focused use case, where Watson is integrated to help customers choose which combination of fifty granola ingredients work best together, as in: what goes well with jalapenos? Phil concludes: "AI plays a crucial but very specific role here in informing the selection of ingredient choices — a pragmatic use of the technology." These modest use cases may lack sex appeal, but I like how they don't overreach. In each case, AI is integrated with a range of cloud services to provide an outcome. I just wish I ate granola.
diginomica four - my top four stories on diginomica this week- International Women's Day 2017 - a round-up of tech's progress - March 8 was International Women's Day. Good time for Madeline to assess the progress. I get why Madeline finds the need for a "women's day" depressing; the declining numbers of women in tech isn't feel-good news either. But, Madeline also calls attention to a number of worthy initiatives, such as #techmums and Lego's women in STEM push.
- Reap what you sow – talent shortages or a failure to plan training? - Brian is singing my song, baby. He's fed up with the "talent shortage" excuse when too many companies fall short on training. Training can be perceived as dull documentation. Done right, it's about mentoring, and a talent advantage. Or as Brian puts it: "The best teachers and content may already exist within your own firm. Are you doing everything you can to get their knowledge and experience transferred to as many people as possible?"
- Costco - an e-commerce tortoise takes on the omni-channel hares - Nice to see an unapologetic e-commerce tortoise like Costco with open disdain for the omni-channel "hares." It's all about the in-store customer for Costco, but, as Stuart notes, that doesn't mean Costco isn't spending on tech. Hard not to root for the tortoise, who won the race last time we checked. For a different twist, check Stuart's Morrisons - signs of omni-channel life?
- Keeping pace with digital transformation? It’s a long hard road - I'm not sure if I've ever been alarmed by a PwC study before. This one came close. Barb examines the findings - and asks why companies' so-called "digital IQ" seems to be going down, not up. One potential reason: technical advances are moving faster than companies are.
Vendor analysis, diginomica style. Here's my three top choices from our vendor coverage:
- Exclusive Interview - Passing the CEO baton at ServiceNow: Slootman to Donahoe - Derek gets the dual interview at the moment of handoff: "ServiceNow believes that Donahoe’s tenure at eBay, a consumer-focused company, will bring valuable experience to the enterprise software company. Slootman said that the number one conversation he has with CIOs nowadays is no longer about cost, which is what it used to be, but about the need to bring the enterprise experience in line with the consumer experience."
- Watson, meet Einstein - a relatively elementary AI alliance - Smart bedfellows? Stuart with more on this meeting of the AI minds: "For example, Watson could analyze shopping patterns, weather and retail data, working in conjunction with Salesforce Einstein to help a retailer send automated personalized marketing emails to customers. Or an insurance company using Salesforce could use Watson’s weather data to target customers ahead of storms knowing that they will want to protect their property."
- Inside SAP's Google Cloud partnership news - a Google Cloud Next review - Brian Sommer and I join forces for our initial take on a cloudy partnership, with only a smidge of snark. Why now? "This relationship is a result of changing IT tastes. Public cloud solutions are on the upswing as are the improved economics that result from them."
A few more vendor picks, without the quotables:
- Kiviks Musteri anticipates bumper harvest from shift to Infor cloud - Jessica
- Why the Microsoft Gods may own the Mount Olympus of how data centers develop - Martin
- Dynatrace brings AI smarts to APM as enterprises transform - Phil
Best of the rest
The problem of defining AI, and managing AI projects by several smart peepsquotage: "Explaining what a machine learning model does in business terms is a non trivial challenge to begin with . Now to explain why the model doesn’t work as planned and why it needs to be reworked – without guarantees it will work the next time – makes it a much harder problem . Even when things work well – and if a business person asked you to explain why the model arrived at the result it did , its often hard to explain." - Vijay Vijayasankar, So, you want to manage AI projects?
myPOV: As we get further along the AI/machine learning hype adventure, field lessons bring a welcome gut check. That's the case in Vijayasankar's post on managing AI projects, where he isolates the project management issues by assuming that the AI/algorithmic services are provided for. Even so, we don't get far, tripping up on standard phases like project estimation: "Whatever we estimate – there is a good chance that while developing we will need to switch drastically to something else and that will invalidate prior estimates."
Then we get to the next issue: "Your source data can fundamentally change the fate of an AI project." But hold up. Isn't ETL/data cleansing always an issue on enterprise projects? Yes, says Vijayasankar, but even if your AI data is clean, your model may not find the data useful. And so it goes. Some may find these lessons a cold shower, but we need these field learnings.
This one isn't field lessons, but it's got a practical bent: Artificial Intelligence Will Turn Journalists Into "Centaurs," Not Replace Them. Media tech expert Stephen Masiclat says we're going to become Centaurs. I'm not sure I want the body of a horse at this time, but Masiclat is talking about a fusion into something new, a human/AI collaboration combo. Perhaps the blogger's saving grace: AI can't understand the emotional elements of a story. Then again, how many of us do? Finally, for more practical definitions check The difference between vertical AI and general AI (nutshell: "vertical AI" is a good match with this week's practical AI themes).
Google Next review: Constellation was all over Google Cloud Next, with Holger Mueller and Ray Wang each shooting impromptu videos in their tarmac-teaser style. The themes? Google's push for enterprise cloud legitimacy. Me: agreed on the progress, but Google isn't there yet. The developer/platform play is promising, though. Wang's post: Event Report: #GoogleNext17 On Path To Enterprise Ready. Mueller: Event Report - Google Next 2017 - Google makes progress - but is it fast enough?
Honorable mention
- I invented the web. Here are three things we need to change to save it - One of the founding fathers doesn't like what he sees - hard to blame him. FYI - I addressed several data privacy related pieces today in Consumer privacy is a hot mess - enterprises can do better.
- Down Report - Human error takes AWS S3 down in US-EAST-1 - and it is felt - 3.8 Cloud Load Toads - The longest, most
self-indulgent noodlycreative post title of the week is claimed by Holger Mueller. In this PR deconstruction, he raises points on Amazon's accountability I haven't seen before. - Don’t Start a Customer Relationship with a Lie - A rant with a purpose, targeting the friction many companies jam into their so-called "free trials."
- C.I.A. Scrambles to Contain Damage From WikiLeaks Documents - A bad vibes story all around, raising tech consumer privacy questions while adding fuel to the "could we be next?" leak worries.
- Bend me, shape me: cloud applications not always as agile as hoped, survey suggests - as customers get more cloud experience, their complaints mount also (inflexibility being high on list). Not sure "low code" is the way forward, I'd argue for platform-based extensibility.
Whiffs
Hopefully this isn't fakery, but it looks like "The Real Hunger Games" is gearing up for web reality TV from a "bear-infested island in Siberia." The show's producers say they won't intervene even in the case of violent crimes (though authorities might). Elsewhere in human suffering, Game of Thrones producers made fans watch a block of ice melt for an hour to learn the premier date for season 7.
I don't normally embed videos, but the home interview misadventures of this unfortunate professor on BBC is an all-timer:
Oh, and Google Home is getting a nice head start in the fake news game, serving up wrong and inappropriate verbal queries. The crud commercials on my local sports network are really starting to bug me. So last week, I provided a helpful ranking of the worst ones:
.@CSNNE annoying ad rank::
1. @legalseafoods - unfunny obits
2. @Amica - "helpfulness insured" lol
3. @LaheyHealth - phony personalization— Jon Reed (@jonerp) February 28, 2017
Surprisingly - no response. On the enterprise side, is this the most over-jargoned, unappealing blog post title of the year? You tell me: Digital Business Distributed Business and Technology Models Part One; Understanding the Business Operating Model. We all struggle to talk about tech in plain language. But - not much sign of a struggle here I'm afraid.
Over to you, Clive.
Which #ensw pieces of merit did I miss? Let us know in the comments.
Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed. 'myPOV' is borrowed with reluctant permission from the ubiquitous Ray Wang.