Artificial Intelligence (AI) in the Enterprise

Advertisements
Image | Posted on by | Tagged | Leave a comment

How Companies Are Using IoT Now

IoT_Use.jpg

Source: ZDNet

Posted in Internet of Things | Tagged | Leave a comment

Top Use of AI in the Enterprise Today is Assisting with IT Activities

Source: How Companies Are Already Using AI

Posted in AI, AI Enterprise | Tagged | Leave a comment

Data is Eating the World: 163 Trillion Gigabytes Will Be Created in 2025

idc_global_annual_datasphere_size

Source: Data Age 2025

Posted in Big Data Analytics, Data growth | Tagged | 1 Comment

Will Google Own AI? (5)

AI_GitHubCommits_DeepLearning

AI_GitHubContributors_DeepLearning.png

Matthew Rubashkin, Silicon Valley Data Science:

“[Google’s] TensorFlow is the 800-pound Gorilla in the room in regards to quantity of tutorials, training materials, and community of developers and users.”

Mentions of Deep Learning frameworks in papers uploaded to Arxiv-Sanity in March 2017:

% of papers 	 framework 	 has been around for (months)
------------------------------------------------------------
    9.1          tensorflow       16
    7.1               caffe       37
    4.6              theano       54
    3.3               torch       37
    2.5               keras       19
    1.7          matconvnet       26
    1.2             lasagne       23
    0.5             chainer       16
    0.3               mxnet       17
    0.3                cntk       13
    0.2             pytorch       1
    0.1      deeplearning4j       14


Andrej Karpathy:

10% of all papers submitted in March 2017 mention TensorFlow. Of course, not every paper declares the framework used, but if we assume that papers declare the framework with some fixed random probability independent of the framework, then it looks like about 40% of the community is currently using TensorFlow (or a bit more, if you count Keras with the TF backend).

See also

Will Google Own AI?

Will Google Own AI? (2)

Will Google Own AI? (3)

Will Google Own AI? (4)

 

Posted in AI, deep learning | Tagged | Leave a comment

Google Enlists AI and Community Development to Win the Cloud Wars

Fei-Fei Li, Google

Reflecting the rapidly increasing interest and investment in cloud computing, 10,000 developers, engineers, IT executives, and Google employees and partners gathered at Next ’17, Google’s annual cloud event for enterprise customers. Google showcased customer testimonials from Disney, Verizon, HSBC, Colgate-Palmolive, and Ebay; support from a number of new partners, including SAP; and a series of AI and cloud infrastructure-related announcements. (Disclosure: Google covered most of my travel expenses).

Analysts were not impressed (see here and here, for example). Google, the argument goes, is a consumer company and does not understand or have ready access to enterprise customers. Amazon’s first mover advantage has produced an insurmountable market lead (last year, it reported cloud revenues of $12.22 billion, compared to Microsoft’s $2.42 billion and Google’s $900 million, Deutsche Bank estimates). And other companies with a large footprint and years of experience in the enterprise market have staked their survival on capturing a large piece of the cloud (see IBM and Oracle). Google is late to the cloud party, is doomed to remain a minor player, and may even end up nixing the whole effort as it has been fond of doing with less-then-spectacularly-successful applications and initiatives.

Excellent arguments all but I beg to differ. Google stands a very good chance of winning big in the cloud computing market, possibly gaining top market share by 2025. It will be an uphill battle but Google will succeed provided it executes flawlessly along the two lines of attack where it is already king of the hill: Artificial Intelligence (AI) and community development (aka “openness”). Succeeding in the cloud computing market will be the natural evolution of what Google has unleashed, quite accidently: The data-centric transformation of the entire IT market and of the business world.

“Google has been a machine learning company for a long time, every one of its consumer-facing products has been powered by machine learning,” Fei-Fei Li told me on the sidelines of the event. She continued: “Google gets it. It has a deep bench of AI and machine learning technologists. It’s clearly in our DNA. Now, we see the opportunity in the enterprise world.”

Fei-Fei Li and Jia Li, both leading experts in AI (specifically, in computer vision), joined Google last November to lead its newly created Cloud AI and Machine Learning group. “Ten years ago,” said Jia Li, “people told me no one needs image understanding or image search. Today, industry is much more aware of what machine learning can do. Now there are so many new applications for image search.”

To ensure a steady flow of new AI applications, they want to “democratize AI,” lowering the barriers of entry to AI. Their mission is to “make it available to the largest community of developers, users and enterprises, so they can apply it to their own unique needs.” In her keynote, Fei-Fei Li broke down the mission into four elements: Computing, algorithms, data, and talent. It’s the combination of the processing power and geographical reach of the cloud; the availability of Google’s machine learning models and APIs; sharing with the world massive data sets; and transferring Google’s AI skills and expertise to enterprise customers.

The last two components were behind Google’s acquisition of Kaggle, a startup established seven years ago “to make data science a sport.” It is a community of more than 800,000 data scientists and hundreds of high quality open data sets. Kaggle sits “at the nexus of many things we care about,” says Fei-Fei Li. Jia Li adds: “The Kaggle community has expertise in a lot of machine learning problems from many different perspectives. They can help us reach the larger data science community.”

Probably more than any other entity, Kaggle has been instrumental in making data science “the sexiest job of the 21st century,” including ranking data scientists by their sex-appeal, i.e., according to their performance in Kaggle’s competitions. As these data scientists are independent of Kaggle, in acquiring the company Google may have simply wanted to ensure the Kaggle platform for data science competitions and data sets sharing will continue to exist, rather than falling into the hands of a competitor. A more direct control over the community will also help Google in bringing top machine learning experts into its AI ecosystem.

Developing a large ecosystem or a dedicated community of followers is also in the “Google DNA” and will serve as an important weapon in the cloud wars. This was used to be called “openness” in the IT industry, but Google (and other companies of its generation) had taken it to a whole new level.

In the old IT industry (i.e. in the 1980s and 1990s), to distinguish your products from the likes of IBM and its customer “lock-in,” you claimed your own products were “open,” conforming to “standards” that allowed customers to easily move from one vendor to another. Unix, and later, Linux, were the standard bearers of the “open” movement. They demonstrated that what “openness” actually meant was developing a community of loyal followers.

Google took this concept to a whole new level, having in its DNA an academic bent, a “publish or perish” mentality. That resulted in publishing a string of papers providing details of its internal IT innovations and the sharing of code on popular open source platforms such as GitHub.

As in the old IT days, “openness” has implied more self-interest than altruism, but it has worked in new ways to bolster Google’s standing in the IT community, long before it finally decided to sell cloud computing to that particular community.

Here are some of the ways by which “openness” or community development works for Google:

  • Even the best programmers may take months to learn Google’s internal systems and become productive. But if they have already used Google’s open source code, they are highly productive the moment they are hired.
  • Google can tap into the minds of the best programmers in the world (besides those already working for Google) to improve its software.
  • Given the public nature of the platforms for sharing (and commenting on and adding to) open code, Google can tell who are the top (and most involved) programmers and hire them.

Basically, “openness” is network effects applied to people, not technology—it’s best to know what everybody else knows. By sharing its IT expertise with the world, Google, the consumer-facing company, has become a key influencer in the IT world, albeit somewhat inadvertently. It was not alone in becoming an IT influencer, as other companies born on the Web (e.g., Amazon, Netflix), had also faced unprecedented requirements for uptime, processing power and storage capacity, demand spikes and security threats. In the process of responding to increasingly demanding online and mobile consumers, these companies have initiated what has become to be known as “the consumerization of IT,” exemplified (in the most simplistic way) by CEOs asking their CIOs, “why can’t our IT systems be as simple to use as the hardware and software I use at home?”

After the dot-com bust in 2001, those CEOs and CIOs of traditional companies, “breathed a sigh of relief—that Internet thing didn’t happen, we don’t need to worry about it,” recounted Marc Andreessen at the Next ’17 event, in a conversation with Vint Cerf moderated by Quentin Hardy. They maintained that view until 2010, said Andreessen, when they woke up to the promise of cloud computing. Or maybe these weren’t the same executives, as the shift to the cloud has also coincided with a generational change, argued Andreessen.

I remember the widespread belief in the early 2000s that “IT doesn’t matter.” It was a failure of imagination, mistakenly confusing the hype about “the new economy” and the dot-com bust with the fate of IT, leading to the conviction that IT was not any more a source for competitive advantage and the revival of the (1960s) notion of the “computing utility.” Exactly the opposite has happened and it turned out that IT matters a lot and that providing it from the cloud generates profits no utility has ever seen.

Until about 2014, said Andreessen, a popular Wall Street theory has been that “cloud has to be a commodity” and everyone told him that it will be impossible for a cloud vendor to make money. Then Amazon started to report separately the results for AWS (its cloud platform), and it turned out that it “has by far the best margins in Amazon.”

Because IT matters more than ever, the work of IT professionals has expanded to include many new activities and types of jobs, interesting and highly valued tasks that directly impact the business. The more mundane work, the management and maintenance of the IT infrastructure, many companies have recently concluded, is best left to the cloud vendors. As Alphabet Chairman Eric Schmidt asked rhetorically at the Next ’17 event, why would any business want to replicate Google’s $30 billion investment in its IT infrastructure? “Leave the infrastructure to us,” was his advice.

IT matters a lot because it has become a key foundation for business success, starting with companies like Google that were propelled by the Web and its unleashing of a new and spectacular explosion of digital data. This data tsunami is now taking over all types of companies in all industries and to handle it, they need to rely on the cloud for its processing, storage, sharing, mining, and driving AI tools.

The period between the invention of the PC and the launching of the Web “completed the first half of the data story,” said Fei-Fei Li. “The second half of the story is data analytics and intelligent machines.” In her opinion, data is the essence of the 4th industrial revolution.

Whatever label we put on it, what we see today is a worldwide movement to find new opportunities, insights, and solutions in data. Lots and lots of data, served by a secure and reliable IT infrastructure managed by cloud vendors. “In the last 6 years,” said Andreessen, “we have seen Silicon Valley companies using [cloud computing]. Now we see a new generation of startups use this technology to go after industries where tech was not important.” In her keynote, Diane Greene, the head of Google’s cloud business, said that what Google offers is “technology built for the enterprises we have today or starting to have.” In other words, the technology developed to serve the needs of a data-centric enterprise such as Google, is the perfect solution to the requirements of all enterprises which are now becoming more and more data-centric, collecting and keeping data on everything and finding new opportunities to use it for improved operations or new revenue streams.

But data today is more than just a business catalyst. Andreessen talked about how data provided by companies such as Yelp and Trip Advisor has fundamentally changed how consumers make purchase decisions—and Hardy added that it also changes their tastes and preferences. Indeed, one might coin a phrase: “Data is eating the world.”

In eating the world, data has not only transformed the management of IT and the IT industry, it has also blurred previously rigid industry boundaries and destroyed the sharp distinction between what is considered “consumer” and what is considered “enterprise.” When everything looks like ones and zeros and you focus on collecting and mining as many ones and zeros as possible, old categories just fade away.

What are the chances of success for Google, the “consumer” company, in the “enterprise” market? Quite high, I would argue, since the requirements of the enterprise market are now where Google was in its internal IT journey about a decade ago. It has about 1,000 machine learning models in production; has trained 10,000 Google employees in machine learning best practices; has best-in-class experience and expertise in AI; has written the book on running online operations without a hitch and has made this expertise available to its cloud customers; and so on.

Of course, Google faces a formidable competition which will only increase as more and more enterprises will move their internal IT operations to the cloud. This may include new competitors such as Facebook—it has developed (with the help of some ex-Googlers) an IT infrastructure serving close to 2 billion users and has shared with the world its IT expertise in the Open Compute Project. At some time in the future, will Facebook appoint a cloud-savvy member of its board of directors (e.g., Marc Andreessen) to head its new cloud platform business, just as Google did two years ago with its cloud-savvy board member Diane Greene?

At Next ’17, Google’s CEO Sundar Pichai called the massive knowledge transfer from Google to enterprises “an extraordinary bet for us” and said that “bringing our infrastructure to serve the needs of businesses worldwide” is “a natural extension of our mission [to organize the world’s information].”

Well, I could have told Pichai a few years ago that Google should tweak its mission to something like “making the world’s data useful” and change its business model accordingly. As a matter of fact, I did. In August 2011, I wrote:

So far [Google] has let Amazon take the lead in encroaching on traditional IT vendors’ territory.  Why?

In the interview [with Mossberg and Swisher regarding what he called the “gang of four”—Google, Facebook, Apple, and Amazon], Schmidt was asked why he left Microsoft out… His answer: “Because Microsoft is not driving the consumer revolution in the minds of the consumers. Microsoft has done a very good job of getting itself locked into corporations and much of their profits now comes from the union of Windows server and the clients, which they do very well at.”

Consumer vs. enterprise is an old and soon-to-be obsolete distinction. If Google will not take away some of Microsoft’s (and IBM’s, etc. for that matter) “enterprise” revenues, someone else will.

At stake are the $1.5 trillion spent annually by enterprises on hardware, software, and services. If you include what enterprises spend on IT internally (staff, etc.), you get at least $3 trillion. A big chunk of that will move to the cloud over the next fifteen years.

Compare this $3 trillion to the $400 billion spent annually on all types of advertising worldwide.  Why leave money on the table?

In 2011, Google got more than 90% of its revenues from advertising. Today, advertising still accounts for 85.9% of Google’s revenues. Transferring its IT expertise and data mining and machine learning knowledge to enterprise customers will finally make it less dependent on a slow-growing advertising market.

For Google, the cloud is the future. On the first day of the Next ’17 event, the San Francisco Chronicle reported that the Mountain View City Council approved Google’s “futuristic new campus,” the first major real estate project that Google will build from the ground up: “At its heart is a planned 595,000-square-foot, two-story office building shaped to resemble a puffy white cloud.”

Originally published on Forbes.com

 

Posted in AI, Google | 1 Comment

Logz.io Combines Open Source, Cloud, Big Data and Machine Learning for DevOps and SRE

cognitive insights screenshot

Logz.io Cognitive Insights

90% of startups fail. Those that survive and thrive either capitalize on a new technology or provide a timely response to a new market development. Logz.io does both, combining the power of at least 4 technology trends—open source, cloud computing, big data analytics, and machine learning—while addressing a new group of influencers in IT purchasing—DevOps staff and Site Reliability Engineers (SREs).

Logz tells its customers what’s going on with their software applications. It offers an enhanced version of the open source ELK stack which combines an enterprise search engine with log analytics and visualization tools. On top of ELK, it has developed Cognitive Insights, an artificial intelligence platform that detects overlooked and critical events and provides the user with actionable data about context, severity, relevance, and recommended next steps.

“Log data is very noisy, very susceptible to false positives,” says Tomer Levy, co-founder and CEO of Logz. To figure out when and where its customers should pay attention, Logz combine in one very large database in the cloud (i.e., Amazon Web Services cloud) data from 3 sources: What it finds by searching the relevant online discussion forums where software developers talk about specific issues and their solutions; what it learns from the experience of its customers; and what it learns from how its customers have reacted to specific events.

In other words, Logz crowdsources expert knowledge from the best engineers in the world and then applies it in real time to offer recommendations to its customers as critical events happen. “We take communal knowledge and run it on your data,” says Levy. Without the type of assistance they get from Logz, software engineers can rely only on their own limited knowledge and experience and typically just guess at what needs to be done.

With today’s complex, massive, and distributed IT infrastructures, it is close to impossible to find the correlations and relations between specific occurrences of the same piece of information in the vast amounts of log data. At this scale, you need artificial intelligence to observe, learn and provide recommendations based on the experiences of hundreds of software development teams and hundreds of thousands of contributors to discussion forums.

Open software and open discussions have fueled the rise of “digital natives” such as Google, Facebook, and Amazon. These companies, in turn, have contributed greatly to all the technological developments Logz is capitalizing on—developing in-house new software applications to allow them to manage IT operations at unprecedented scale and then sharing these applications with the world as open software; building a massive, distributed IT infrastructure, and in some cases, providing it as a cloud service to other organizations; collecting, analyzing and running their companies on big data; and using machine learning to analyze the data and advance every aspect of their business. In the process, they have also created new IT-related roles and changed how IT solutions are acquired.

The IT world used to be neatly divided between software developers responsible for developing applications and operations staff responsible for running and maintaining the applications in production. But this division of labor did not work for the online life of digital natives. Specifically, early in Google’s corporate life, its new practice of constantly revising its software and continuously introducing new features, while at the same time maintaining high quality user experience, required a new approach. The solution was the creation of a new team, Site Reliability Engineering, applying the disciplines and practices of software engineering to managing IT operations and running production environments.

The marriage of software engineering with IT operations in the form of the new role of the Site Reliability Engineer spread quickly to other high-tech companies. But about ten years ago, the concept was taken a step further in the form of the new DevOps movement. As all businesses are becoming digital businesses, they are also adopting the practice of Agile software development, continuously revising their software applications. They also need to answer the new requirements of employees, customers, and partners expecting an always-on, secure, and reliable IT infrastructure.

Today, IT operations has become part of software development. For applications to be resilient and fault tolerant, they need to include constant monitoring, adaptability to large swings in load, and Web performance optimization. The DevOps team takes care of these new requirements, now at a growing number of organizations, of all sizes and in all industries. And they have an increasingly significant role in evaluating and recommending tools that ensure the new IT infrastructure is always humming, always performing.

“DevOps is about accountability,” says Levy, “not just writing the code but also overseeing how the code is running in production.” The DevOps team is where business users now go to complain if they think there is a problem with how the application is working. This new development in the IT market has created an opportunity for Logz (and many other startups) to address a new influential group that can give them an entry into the business.

“We launched the product in October 2015, we have raised $24 million to date, and now there are more than 1000 companies in 80 different countries that use Logz every day,” says Levy. While he describes the product as a “DevOps solution,” there are a number of different uses for it in addition to application monitoring, including business intelligence, compliance, and security, the latter now accounting for 20% of revenues. For example, Rent-A-Center has reported that Cognitive Insights helped it detect potential security threats before they impact its customers. And Logz helped Dyn quickly respond to and mitigate the unprecedented DDoS attacks it recently experienced, attacks that temporarily shut down or slowed many popular websites.

Other brand names Logz.io counts among its customers include British Airways, CNN, Electronic Arts, and Oracle. Sitting at the confluence of the latest technology trends and being propelled by a rising group of IT influencers is a proven recipe for startup success.

Originally published on Forbes.com

Posted in AI, Big Data Analytics, Cloud Computing, Machine Learning | Tagged | Leave a comment