In the last five years, AI has become a buzzword that spurred a lot of debate and discussions. We live in an increasingly digital world and continue to rely more and more on digital in almost all aspects of our lives. So, there is a need for us to understand the basic concepts and differences regarding that field. However, some of the first uses of AI technology is found right there in the IT sector.
Those working in the IT sector are acquainted with the concept, but what about enterprises and businesses who collaborate with an IT team or have an in-house IT department? Do you know what Artificial Intelligence (AI) is? What about Machine Learning (ML) or Automation in general?
We are bringing to you a summary of a recent ONUG town hall meeting, moderated by Neal Secher (SVP Head of Networks & DC Modernization at State Street), along with Bob Friday (CTO & Co-founder at Mist), Viraj Parekh (Executive Director – Global Product at Verizon), Prakash Seshadri (Software Engineering Director at Juniper Networks), and Mike Albano (WiFi Operations Engineer at Google). They discussed the recent developments in AI and its positive impact on the wireless networking space.
Take a few minutes to read about how AI is going to re-define the networking space and influence the digital transformation of the business landscape, according to these IT experts and thought leaders.
The Differences between AI, ML, and Automation in General
Before proceeding to the topic, the participants laid some groundwork on the fundamental differences between AI, ML, and Automation in general. We can define AI as the awareness ability of humans and the processing power of machines coming together. There is no AI without ML, and there is no ML without data science. The main questions that drive their research are whether we can create something that can do a human job with a certain level of intelligence.
As for the current state of application of AI in IT, we’re still far away from having completely self-driven systems that can allow IT experts to take their hands off their networks. However, they say that we are on the right path. And the journey begins with automation, which is the starting point on the infrastructure and network side. It all comes down to automating the sequence of steps based on different trigger points to achieve a specific outcome. With ML, you can learn, adapt, and self-program using the available processing capabilities, which can further take automation to a whole different level and context. Also, Automated Machine Learning brings the ability to choose different data sets and find relationships which we didn’t know about, creating actionable insights at the infrastructure and network level. Thanks to the computing power, we will be able to do more than a single human could ever imagine doing.
AI in IT: Where Are We Today?
What do we see that’s actionable in the IT environment (from an AI perspective)?
Bob Friday explains that they’re on a mission to build a solution that can give answers on par with foremost networking experts (they’ve released a system that can do that, but it’s still on a “rookie” level.) The system is there to make things easier for an expert team to get more accurate answers quickly.
With an automated system governed by AI, you’ll have a centralized “person of assistance” that can correlate different events and provide an intelligent response, which is a current trend that Prakash Seshadri believes is going to explode (based on various use cases). To achieve the best answer, IT experts need to correlate different systems, and this will be a stepping stone for what operators have been doing this whole time.
According to Mike Albano, the connectivity with cloud providers is unique as they all have slightly different networking models. There are no standard approaches to cloud connectivity which makes it more difficult. Standards are necessary, so what is this AI going to look like?
What Are the Initial Building Blocks When Selecting an Infrastructure?
Mike Albano discusses that we need the standards scheme of APIs among vendors and multiple sets of equipment to bridge the gap for those who want to make the AI. Real-time cloud stack is also a vital component of a self-driving network. Bob Friday also adds that good AI starts with useful data. The fundamentals towards which these experts are going for having visibility to the data, being able to collect it, create actionable insights, and ultimately take those actions.
Where to Get the Data to Run the Algorithm?
It is where interoperability comes into play. Standardization of domain-specific intelligence (depending on your business) is required to make the collaboration between vendors easier and less costly. There needs to be standard data formats to get normalized data that can be collected and exposed in a way that can be shared with partners.
Will Companies Have to Buy Third-Party Systems or Build Them Themselves?
There is room for both options, for a kind of a hybrid model. When it comes to domain expertise, enterprises will have to depend on vendors. On the other hand, the business requirements are where you need to bring your knowledge to the table.
These experts are all working on the critical segments of AI systems that can be applied in IT. They envision a kind of “personal assistance” that can correlate many different events and provide an accurate and intelligent answer, faster and more precise than human resources could. However, before being capable of building that system to a high level, there must be some great groundwork laid regarding standardization and increased interoperability.