MyPOV: AI had a chance to learn from the mistakes of tech past, but as Stuart reports in Tackling bias in an AI sector that’s already too ‘pale, male and stale’, not so fast, people. Stuart assesses a new study from New York University’s AI Now Institute, and finds that "AI has a diversity crisis well underway."
Though Katia Walsh is breaking the mold at Levi Strauss in her new role as Chief Artificial Intelligence Officer, she's the bold exception, not the rule. The report's concerns go deeper. Stuart warns:
This isn’t just a problem of recruitment and talent management, but one which has potentially problematic knock-on effects when it comes to elimination of bias in AI systems themselves.
Neil continues the deconstruction with The blush is off the rose of Machine Learning…maybe. Yep, bias is everywhere:
We already know there are drawbacks and even dangers with Machine Learning (ML). The literature is replete with examples of bias, not necessarily from the algorithms themselves, but from biased datasets, bias in selecting training data and all sorts of insidious bias that is picked up in the data, hiding in plain sight.
But Neil stirs the pot with a new ML problem - the geeky/necessary problem of performance:
A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models.
(Ouch!). Neil then
geeks out parses another deep learning study, concluding that "the extreme cost and effort of ML, especially Deep Learning approaches like ANN, as opposed to the tried-and-true LR, which every statistician understands," is not justified by this data. Neil is still bullish on AI, but not on its idealization:
AI will be the most important thing to ever happen to technology, but not this week.
Diginomica picks - my top stories on diginomica this week
- Bringing cloud services to the edge - We keep our geek on with Kurt's assessment of further progress in the (mega-buzzword alert!) edge cloud: "the notion of an Edge cloud built from micro data centers becomes compelling." Why? Everything from the impact of 5G to the surge in streaming video points to the edge.
- Omada Health uses digital smarts to ease the fight against chronic disease - Michelle filed a nifty Zendesk use case, where behavioral science, connected technology and customer service converge in the fight against chronic disease.
Vendor analysis, diginomica style. Here's my three top choices from our vendor coverage:
- Workday maps out an AI-powered 21C enterprise architecture - Phil updates from the Workday Analyst Summit. I was interested in the "self-modifying user experience." I also get Phil's concerns about helping end users cope with the pace of technical change.
- Huawei Analyst Summit – former UK Government CIO says, ‘Put us to the test’ - Huawei in the unenviable position of being accused of things that are difficult to prove - or disprove. Martin shares how they are turning this controversy "into a feature that its products are far more thoroughly and comprehensively tested than any other comms equipment available."
- SAP earnings Q1 2019 – accounting for Qualtrics and restructuring, as an activist investor takes a stake - SAP's Q1 earnings news was anything but routine, bringing with it loaded news, including the separate announcement of an activist investor. Here’s my pre-Sapphire Now review.
A few more vendor picks, without the quotables:
- Microsoft provides a reminder that AI ethics is not a question of black and white - Stuart
- Demystifying Zoho - CEO Sridhar Vembu reveals Zoho's ambitions in a post-ERP world - Jon
- Freshworks CEO - “What if there’s an iPhone moment in business software?” - Derek
Jon's grab bag - Stuart always saves some vinegar for two
misbehaving delinquent overreaching companies that are frequently called into the principal's office just like you and me probably were in high school. Start with: A $5 billion privacy fine is a rounding-up error for Facebook, not a reason to change. Tell us how you really feel Stuart:
There’s not much to add here other than that personally I don’t believe a word of this honed and scripted strategic repentance.
Twitter doesn't escape Stuart's analytical barbs either: Twitter, trolls and Trump – Jack Dorsey’s busy day. All I can say is: you have low expectations when it comes to making gutless anonymous dweebs feel unwelcome, Mr. Dorsey. Still, Stuart cuts Dorsey some slack:
The clampdown on abuse and fake accounts clearly had some proof points.
Let's hope so, otherwise, Dorsey's
daffy doodles brilliant ideas for radically changing Twitter might resurface.
Best of the rest
Lead story - What Is Digital Transformation, and How Do We Get There? by Frank Scavo
MyPOV: With all the
belly aching wrangling over the over-used buzzphrase of digital transformation - which I have been known to indulge in - it's refreshing to get such a practical and well-considered definition of digital transformation from Strativa's Frank Scavo:
Digital transformation is the use of digital technologies to enable an organization to improve or evolve into something new.
Scavo goes on to define five categories of digital transformation:
- transforming business processes, relationships, and insights
- enabling new products/services and business models
He makes the important point that these are ongoing, business-driven initiatives, not discrete IT projects. But it's Scavo's first statement that drives home the paradox. Properly applied, the fancy-sounding term of "digital transformation" should be broadly accessible:
Digital transformation should be something practical and tangible, something in reach of all organizations—provided business leaders make a sustained effort.
I would add three points:
- Digital transformation almost always extends beyond enterprise walls. Even internal transformation projects should usually have an external stakeholder group (customers, suppliers, app developers, potential new customers) firmly in mind.
- In past decades, attempts to "extend" projects to stakeholders outside the enterprise ran into deep technical barriers. Those barriers have lessened considerably, via mobile/cloud/APIs etc.
- Successful digital transformation is ultimately enterprise-wide and culturally central, with buy-in on every level of the company. But if you can't launch your first digital projects quickly and affordably to gain traction, you're probably on the wrong track.
Kudos to Scavo for such a terrific digital transformation framework.
- Survey Shows a Security Conundrum - How's this for a recipe for continued breaches, via Dark Reading: "85% of companies rely heavily on employees as part of their security defense. 40% consider employees are the sole "last line of defense" for the company - but 40% say employees are a significant source of vulnerabilities."
- Robots threaten middle-aged workers the most (that's anyone over 21) - Been a while since we had this type of "machines are coming!" alarmist report data to chew on. Still, some important points to consider on the role of automation versus labor stagnation.
- Hertz Takes Accenture Out of the Driver’s Seat - UpperEdge asks pointed questions about a new lawsuit from Hertz, alleging breach of contract against Accenture: "Why did Hertz allow Accenture to step into the product owner’s role for a project that was so customer-facing and brand critical?"
- Reacting to Feedback as CEO - Good advice from Dave Kellogg here, and not just for CEOs either. I like this one: "Remember that “not reacting now” is not the same as “not reacting." This is very important because “not reacting now” is probably the right answer 90% of the time."
- Do we need a better North Star for AI than the Turing Test? - IBM's Vijay Vijayasankar puts his AI project experiences to the (Turing) test. Get this: "A toddler who has seen a picture or two of a dog on her story book can usually understand that the first real dog she sees is similar to a dog she saw in the book. And the odds of her mistaking a dog for a cat are minimal. That’s not true for machine learning at all – it needs a lot of labeled data to get to a comparable level if at all."
The most surprising headline of the week goes to: Austrian prison escapee returns, fed up of life in the sun. The title promised more than the article delivered, but still: There is no shortage of talent. There is a shortage of suckers.
Small victory: Microsoft concedes expiring-password rules are useless. Not a big win, though - not until
user-hostile dorks holdouts like my bank (Peoples United Bank) get the message.
So if you forced yourself to watch over 100 TED talks (in person no less), and the best insights you could glean were
Tony-Robbins-cutting-room-floor-stuff generic insights like "lack of sleep is killing us" and "Social-media platforms are making even big-name celebrities feel insecure," maybe this TED thing isn't really working out.
And... speaking of Tony Robbins, we've got a dandy here:
The Twitter convo sparked a fun/edgy convo about corrupt SEO tactics, but nothing beats Josh Bernoff's final line in Without Bullshit:
Oh, and for the poor sucker who emailed me, here’s your link: Tony Robbins is a fool.
I guess Tony Robbins will be
link-begging emailing me pretty soon too. Hey, we all need something to live for. See you next time... If you find an #ensw piece that qualifies for hits and misses - in a good or bad way - let me know in the comments as Clive (almost) always does. Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed. 'myPOV' is borrowed with reluctant permission from the ubiquitous Ray Wang.