We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
The combination of Wi-Fi 6 and 5G mobility, combined with an increasingly wired and mobile world of internet of things (IoT) technology, promises to bring billions more devices onto networks in the coming years. This will have a profound impact on workplaces of the future, in ways that go far beyond the clear trends of remote employees and hybrid workforces.
The world is entering a place where many people can seamlessly connect with fellow workers virtually from any location, with the workplace becoming more intelligent and hoteling becoming the norm. Examples include the ability to schedule a desk similar to seats at the movies or a flight, as well as the ability to crowdsource the temperature in the office. Further, virtual reality and IoT sensors will allow remote expertise to be brought anywhere in the world.
On the one hand, these new capabilities are exciting. On the other hand, it presents a problem — how will enterprises handle all the massive IT challenges this will present? In addition to having to find the root cause of poor internet connectivity and poor Zoom experiences, IT teams will have to manage thousands of IoT devices with no endpoint security, adding greater complexity to the network environment. And the lack of endpoint security on IoT devices adds a new potential attack surface for hackers.
In the future, the growing capability and scale of AI (artificial intelligence) technology will undoubtedly play an increased role in several areas. AI has become anything and everything, depending on who one is listening to or reading. The immediate reality for the enterprise is that AI is bringing better visibility as data is moved to the cloud, creating better end-user experiences for network users, and helping IT teams deploy, operate and maintain ever-increasing complex and distributed networks. It will also usher in an exciting new on-site era of futuristic corporate campuses and automated factories.
AIops: How AI can power the IT help desk
Cloud AI with an unlimited supply of quality data and basically unlimited compute and storage is enabling an AIops support model that can produce results and predictions with very low false positives – filtering terabytes of networking events into actionable insights that a network team can act on. There is a myriad of such events, from a user logging onto the network to a security camera sensing motion. In transportation, AI is on the verge of driving on par with humans, and in healthcare AI is making diagnoses on par with doctors. Similarly in networking, AI is starting to enable the management and operation of networks on par with human domain expertise. AI is going to be a disruptive force in all industries and all facets of our lives.
Right now, an IT domain expert can ask an AI-enabled virtual assistant to answer a routine question, in plain language, as they would an IT colleague, and get a response with an actionable recommendation and the data to support it.
While this is beginning to happen across the industry, Juniper’s Marvis – an AIops virtual AI assistant based on natural language processing (NLP) and natural language understanding (NLU) technologies – is being used today to proactively mitigate problems in customer networks and help troubleshoot every support ticket. With the current level of technology, Marvis identifies the root cause about 70% of the time with the same effectiveness as a domain expert.
As the technology continues to advance, network operators are starting to trust AI technology like this to proactively discover problems without operator prompts and independently issue support tickets. Within the next five years, Marvis will be playing championship-level networking Jeopardy.
Eventually, this kind of AIops technology will move all the way to the end-user and be a virtual attendee watching over your Zoom call, helping fix bandwidth problems or bad connections in real time. These kinds of scenarios in the enterprise are closer than many think.
In addition to helping IT teams manage and operate networks, AI machine learning (AI/ML) is transforming and improving NLP and conversational interfaces. NLP is an area of AI that is exponentially improving with the introduction of transformers in the 2017 paper “Attention Is All You Need.” From Alexa in the home to Siri on phones, NLP makes it easier for consumers to interact with technology.
In the networking industry, these conversational interfaces are starting to replace dashboards and command-line text interfaces. Equipping AI assistants this way helps IT teams find the root cause to poor internet experiences in an ever-increasingly complex mobile network more quickly and easily. NLP is what gives AI assistants a voice and makes them a trusted member of the IT team.
Challenges in the adoption of AIops
AI and AIops are the next and ultimate step in the evolution of automation in doing tasks on par with human domain experts. Thus, the benefits of AI are well known and increasingly desired by enterprise leaders. In fact, research has shown that 95% of respondents believe their organizations would benefit from embedding AI into daily operations, products and services.
However, while the industry is nearing complete buy-in to incorporate AIops, at the very same time, only 6% of C-level respondents reported that their organizations have adopted AI-powered solutions. So, what’s the rub? Well, the fact is many businesses are standing in their own way when it comes to facilitating successful AI adoption. Most often they fail on one or more of three common challenges — readying technology stacks, preparing the workforce and establishing AI governance.
1. AI-ready technology stacks
Artificial intelligence is only as good as the data it uses to learn – creating, cleaning and managing the datasets and feature engineering remain the biggest technical impediment to widespread AI adoption. Whether it’s not having the necessary expertise on staff to ensure data quality or a lack of computing power, etc., making data ML-ready is a tall order.
This data is based on the continuous monitoring of the network for performance, health and security. Another critical readiness challenge is being able to capture the right data, not just massive amounts of data. And the amount of data can indeed be massive, such as every change in a user state on the network. For example, whenever a user loses connection, domain name service (DNS) or experiences lower throughput.
AI projects often suffer without clear definitions of what is important and what task is being automated. The data journey starts with being clear about what human domain expertise the business is trying to automate and the questions it is trying to answer. So when you start your AIops journey, make sure you are clear on what human behavior you are trying to automate.
2. Readying the workforce
There are three distinct workforce challenges brought upon by the onset of the “AI Era.” That is, organizations need to educate their current employees, recruit from a competitive and shallow pool of highly skilled data scientists and data engineers, and do all of this while answering one of the most difficult questions in IT – “is this new technology going to replace my job or make me more effective?”
Solving the first two challenges involves making the right investments in both training and company culture. There will always be more openings than there are people for highly skilled tech jobs, especially in AI/ML, but if companies can lay the right foundation and consistently invest in training their talent, they will be surprised by just how much they can build on their own. Further, investing in existing employees is a way of reassuring them that AI is here to augment and improve work, not eliminate workers.
Finally, implementing tools and providing opportunities for all employees to apply newly acquired AI skills in their daily workflows helps cement the notion that AI is here to improve their daily experiences. While not every employee needs to learn to code, it’s important to convey that the ability to successfully interact with and leverage AIops can bring immense benefits to many types of positions.
3. AI governance
The data challenge is not limited to the question of how to identify the right data. There is an equally challenging question of what to do with all the data, specifically when it comes to issues of risk, compliance and security. AI invites all sorts of reputational, operational and financial risk, but given the discrete and siloed nature of many projects, those are often unassessed.
There is currently a governance gap in the enterprise, and it’s one of the major threats to stalling AI projects. While the majority of executives agree they have a responsibility to have compliance policies in place, establishing such governance and procedures is often one of their lowest priorities. Organizations can close that gap by bringing in both executive leadership and cross-functional stakeholders to ensure projects that have broad impact are being examined from a company-wide perspective, not just through the lens of one department. Taking this a step further, there is also immense value in appointing AI-specific leaders, as well as formalizing an internal AI center of excellence to ensure governance is given the appropriate level of attention and investment and to support the development of consistent standards across the organization.
While there may be more “hype” around other applications of AI, a key piece often missed is that in the above scenario, the skilled employees who were once spending their days fixing help desk tickets are now able to focus on more strategic activities, application development or some other useful purpose. This automated knowledge management will become the central advance that augments human workers in the enterprise. In turn, this will unleash a cascade of innovation that will have effects throughout every organization.
For example, thanks to the benefits of automating the help desk, network operators and engineers can focus more on synthesizing data into thoughtful decisions that improve a process or service. They will also be freed up to collaborate with the business users of the network, thus making the output of their efforts more strategic. This leads to a reimagining of the enterprise fueled by seamless communications.
Some of this is already coming to fruition in manufacturing with Industry 4.0, which is based on enhanced networking, automation, AI and real-time data leading to smart factories. This removes cost from operations and improves supply chain responsiveness, all while more efficiently meeting customer expectations. Led by AI-enabled network automation, similar benefits will accrue to many other industries.
Bob Friday is the chief AI officer of Juniper Networks.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!
Read More From DataDecisionMakers