Don’t jump into AI without doing your homework

We’ve all seen the hyperbole about the artificial intelligence (AI) revolution and the predictions that it will add trillions of dollars to the global economy in the coming decades. But that all depends on enterprises actually understanding AI and adopting it wisely. A host of companies jumping in unprepared—eager to capitalize on trumped-up promises of superhuman capabilities and market domination—will result in lots of fast money for “AI vendors” followed by disappointment for everyone involved.

So what’s a CIO, or even CEO, tasked with implementing an AI strategy to do? The best place to start is to get up to speed on what AI actually means and doesn’t mean, and what it actually can and cannot do (hint: almost everything you see in the movies or read in clickbait headlines is false). This will help you conceive of useful AI applications within your own company (i.e., what’s easily automatable) and also evaluate the claims of software vendors and SaaS providers that promise AI-powered goodness.

That last point is critical when it comes to saving time and budget. AI vendors is in quotes above because, according to recent research, 40 percent of European companies classified as AI companies actually are not such in any material way. That number seems fair, and those findings very possibly apply to U.S. companies, as well. The right reason to buy a product is because it works, not because it checks the AI box on the strategy checklist.

One useful resource to get started learning from the ground up is the AI for Everyone course on Coursera, which is taught by deep learning pioneer and Stanford professor Andrew Ng (who has also published an “AI Transformation Playbook” for companies). It’s not perfect nor is it applicable to all situations (e.g., it largely overlooks cloud-based AI services), but Ng explains things clearly and provides a fundamental knowledge of how to get from idea to Alexa.

The course is a series of short lectures that first lay out in plain English what terms like AI, machine learning, deep learning and data science actually mean, and what you can realistically expect from investments in those areas today. After that, it delves into how to start building initial AI projects to establish its viability for particular tasks, and what companies need to do in order to successfully put AI to use. Some of those things involve culture and organizational structure, while others are technical—namely, getting your data infrastructure in proper order.

Another option, just announced this week, is Microsoft’s AI Business School. It’s less focused on laying the groundwork for AI than on case studies from Microsoft customers, but that’s still useful because, in the end, AI is really about ideas. And when it comes to finding relevant ideas, it can be hard to beat real-world examples from companies operating in similar spaces. (As a bonus, here’s a good synopsis on how to think about applying AI in the automotive industry for everything from logistics to connected cars.)

To drive the point home, here are some stories—just from the past week—that highlight the risks of betting on AI as a concept without really understanding the technology or how your customers (or employees or whomever) want to use it:

Know yourself and know your audience

Amazon is shutting down all of its US kiosks in favor of expanding its physical stores (The Verge): Say what you will about Amazon’s impact on the retail world, but don’t overlook it’s agile-inspired physical strategy. Starting with mall kiosks was a relatively low-cost, low-effort way to test the demand for brick-and-mortar stores.

Target’s digital sales grew 10X faster than in-store sales in 2018, as retailer adjusts to battle Amazon (GeekWire): You could make the argument that physical retailers going digital actually have an advantage over Amazon, which is trying to do the reverse. But the key is getting the online part right both technologically and culturally, and then fusing it with existing in-store know-how.

To lure young talent, banks mimic tech workspaces (Reuters): The physical layout of, and amenities in, a workspace are definitely part of the allure of tech companies, but experienced developers also expect processes and culture that maximize productivity. This Hacker News discussion on the article provides some useful advice on how to do more than just put up a hip facade.

You can never be too secure

The Marriott breach shows just how inadequate cyber risk disclosures are (Harvard Business Review): A good look at how short-changing security in the name of saving money can end up costing money in the long run. It’s best for companies to take security seriously before class actions pile up and calls for strict government regulation grow louder.

Think privacy’s just a cost center? Think again (Forrester): It turns out that data breaches—not data security—are, in fact, a cost center, so best to protect against them early. And better privacy practices can pay off by inspiring higher trust from customers.

The must-haves for your data center cybersecurity checklist (Data Center Knowledge): Some good advice here on following the NIST cybersecurity framework. These guidelines, developed by the U.S. government, can be helpful in choosing technologies and providers, as well as implementing security protocol.

Can Alphabet become the next big cybersecurity vendor? (Data Center Knowledge): Google parent-company Alphabet’s Chronicle business certainly has institutional knowledge on security at a massive scale, but can it successfully apply knowledge (including via a new log-analysis tool profiled here) to companies in other spaces?

Open source is big, competitive business

On “open” distros, open source, and building a company (Elastic):This is in response to AWS releasing its own Elasticsearch distribution, itself a response to moves by Elastic to lock down some enterprise features. OSS licensing is a whirlwind of change right now, which makes it more important than ever to choose software vendors you trust.

A decade later, Apache Spark still going strong (Datanami): Yes, Spark is still a thing. It’s also an example of how OSS takes different shapes depending on where a project lives and who created it.

Microservices mean responsibility

How data inspires building a scalable, resilient and secure cloud infrastructure at Netflix (Netflix Tech Blog): The data stuff is interesting, but the post starts with what should be the microservices mantra: “Netflix’s engineering culture is predicated on Freedom & Responsibility, the idea that everyone (and every team) at Netflix is entrusted with a core responsibility and they are free to operate with freedom to satisfy their mission.”


Gauge the maturity of your applications portfolio and strategy (Forrester)

Essential skills for successful modern application development (Gartner)

6 things you need to know to safely run Kubernetes
(Pivotal webinar)

My DevOps reading list (Pivotal)

Subscribe Now

Thank you!

Follow us on Twitter Join the conversation on LinkedIn Like us on Facebook Visit our YouTube channel

Pivotal, and the Pivotal logo are registered trademarks or trademarks of Pivotal Software, Inc. in the United States and other countries. All other trademarks used herein are the property of their respective owners.
© 2019 Pivotal Software, Inc. All rights reserved. 875 Howard Street, Fifth Floor, San Francisco, CA 94103. Published in the USA.
Contact Us