By: Matt Fornito, Head of Artificial Intelligence, Trace3
Artificial Intelligence. We hear the term in the media, the news, and in casual conversations. Continuous bombardments with the promise of AI leaves many people pondering, “What is real AI?”, “Are companies actually utilizing AI today?”, and “What happens if we don’t implement AI?” Moreover, beyond the decision of pursuing AI or not – hint: you should – is an equally important question, what is our current level of AI maturity and what is our goal level?
Artificial Intelligence (AI) is an umbrella term encompassing algorithms, methodologies, and various scientific fields. Machine Learning (ML), Deep Learning (DL), Computer Vision (CV) and Natural Language Processing (NLP) are a few of the common fields that fall within this cluster. More simply defined, Artificial Intelligence is the ability for a computer to emulate what a human can do. Or it was. With the advent of big data and companies now owning terabytes/petabytes/exabytes of data, one thing has become clear: humans struggle making actionable insights with large-scale, multi-dimensional data.
AI today is best recognized as a learning agent that excels at detecting patterns and continuously learns and adapts with new data inputs. That is real AI. Leveraging data to make better, more informed, and more reliable decisions. “Data is the new oil” only holds true when that data can help drive innovation. And AI is rapidly driving innovation across the globe.
From retail stores housing mobile carts with built in algorithms that can automatically detect if a customer is wearing their mask or has a fever, to Amazon and Alibaba building fully autonomous robot shipping facilities, to real-time detection of credit card fraud, AI is everywhere. Netflix’s recommender system keeps you glued to your TV by knowing exactly what you would enjoy. Satellite imagery companies use AI to detect installation of unsanctioned nuclear facilities. We now live in a society that is rapidly adapting to AI helping simplify our lives.
In short, with a resounding cry, YES companies are implementing AI. It is estimated that $203 billion will be spent on AI this year and the global GDP will increase $15.7 trillion from AI-enabled enhancements. Analyses indicate half will come from labor productivity improvements and increased consumer demand from said AI enhancements accounting for the rest. Even today, monumental tides are shifting. For instance, the shift from Customer Service agents to virtual agents/chatbots changes the average resolution time for customers across the board from 38 hours to 5.4 minutes with a cost reduction of $15-200 per query down to one dollar.
So, if AI is so important, what now? The first is to ensure AI is important to the business. Your organization needs an internal AI champion to drive digital strategy and innovation. Those who refuse will, sadly, be absorbed by those companies who can reduce costs and increase revenue through AI initiatives.
This leads us to the AI journey. We have developed an assessment that helps you better understand your organization, data, and AI readiness, which is then charted to one of five levels of AI maturity. Let us look at five sample companies at each stage in the AI journey. Please note, that a company’s size or profits is not an indicator of their AI maturity. It is quite possible for a $10 million company to exist at the upper echelon level 5 and a Fortune 50 company to reside at level 1.
Level 1: 5th Grade Recorder
Company A has heard about AI and will sometimes discuss it for fun in the office. Remember offices? Most decision-making stems from past indicators and reports including tableau and excel docs. They hired an undergraduate student with a data science degree. He or she struggles because real world data is messy and the organization lacks statistical experts, a data dictionary, and an AI Champion to promote how AI can drive the business. Most understanding of data is experimental and not put into production. Level 1 is that first instrument you learn – often the recorder. It doesn’t necessarily sound pretty and will require a lot of practice to create something meaningful but produces semblance of something resembling music.
Level 2: Saxophone Ensemble
The organization has chosen an instrument to pursue and a team decided to learn the saxophone together. Harmonies can be created, but there is a lack of diversity in the instruments and thus, the sound as well. Companies at Level 2 start building data science teams – usually about 2-20 members, which may also include data analysts and engineers as a cohesive unit. Some exploratory wins start to occur. This begins to drive continuous demonstration of value; insights are found and pockets of the organization begin to leverage this data science team. The manager of this group is usually the AI champion and can often have difficulty getting the organization to systematically push for AI innovation.
Level 3: Jazz Band
As the organization continues to accelerate around AI and the company starts seeing increased revenue in the ballpark of 6 to 9 figures, leadership begins to support AI from a top-down approach. Infrastructure becomes a larger component here as more data and bigger models require faster networking, faster storage, and GPU-acceleration. Approximately 80 to 90% of models never make it into production, even with verifiable proof of incremental or exponential value of the algorithms. TRACE3 can help drive strategy, recommend hardware builds, and support successful implementations if you want to do more, better and faster. This is where this mesh of guitars, drums, clarinets, saxophones and trumpets start to learn that combining each unique component produces a euphony of sound – a harmonious, melodic tone that makes people truly start to see and listen.
Level 4: Adding the Singer to the Band
If Level 3 is a conglomerate of musicians, Level 4 gives the band a voice. Instead of rhythmic beats, there is now an executive sponsor singing the praises of AI and championing digital innovation strategies for the company. Here, it is integral for enterprise scaling within talent, software, and hardware domains. The team may grow to 20-200 data scientists (or exist in a distributed fashion, in various groups across the company). Software stacks can enable greater success for version control, collaboration, modeling, and production. With so many data scientists, AI-enabled pod architecture is often recommended. One of our recommended partners, NetApp, collaborated with TRACE3, NVIDIA, and Mellanox to create a full stack solution called ONTAP AI. NetApp’s pod enables rapid deployment of AI by providing NVIDIA’s GPU-acceleration, NetApp’s highly-performant flash storage, and Mellanox’s InfiniBand+ethernet switches, driven by the data science and AI expertise of TRACE3 to ensure successful implementation and usage within countless organizations. You’ll be winning a Grammy next year.
Level 5: Symphony
The final level is for the AI whales – those who know their data, know the value of AI hardware, and have implemented AI into production so frequently that the AI team(s) pay for themselves 100 times over. What we truly have is a symphony – and this orchestra represents all the people within the organization building AI, integrating AI, collaborating with the AI team to help their business, or championing AI across the company. The conductor often resides within the C-Suite – many times a Chief Data or Technology Officer, but just as often, the CEO herself. Most or all the company data is leveraged, the high-performance compute and AI workloads are more complex and trying to solve large scale problems.
After reviewing the above, can you identify where your company sits in the Five Stages of AI Maturity? AI is both exciting and challenging. It requires the right people to clean data and build models, the right software to support processes, and the right hardware to drive successful implementation. Most importantly, it requires an internal champion, one who sees the value of AI and wants to build something meaningful. Will you be that champion?
If you want to learn more about how to move through the AI Maturity Model or need guidance on the proper tools, technologies, and infrastructure for successful AI, please reach out to firstname.lastname@example.org and we will gladly sit down with your leadership to educate on AI enablement and implementation.