Demystifying AI for the enterprise at IBM Think

By | 3 minute read | April 9, 2018

AI for the Enterprise, IBM Z Security

Is AI hype, or does it drive real business benefits? What are the main considerations in building an AI strategy that enterprises need to be aware of?

These are some of the questions that I had the privilege of asking a panel of AI experts from IBM, NVIDIA and Gara Guru at the IBM Think 2018 conference in March before an audience of around 400.

At Mark III, I’ve seen it firsthand. AI is such a hot topic for enterprises adopting digital transformation these days that it’s often difficult to understand what is real and what is not. Organizations get started or get further along on their journey from here to AI and can enjoy the realization of enormous tangible business benefits.

Whether it be new digital experiences for customers, innovative new ways to interact with brands, or better predictions and recommendations, AI is already changing the face of industries and identifying the new winners.

What were some of the key insights from the panel that I took away:

  • Deep learning is hot and driving AI: Deep learning is a category within machine learning and AI that employs neural networks to “learn” about data and enable the user to draw new inferences based on what it has learned previously. Are you looking to recognize objects in images, predict risk or translate between languages? Deep learning is most likely at the heart of transformational use cases like these. All of the significant enhancements around AI-focused IT infrastructure and the explosion of data to “teach” these neural networks over the past few years has led to the current level of buzz and the unprecedented outlook.
  • AI systems require a lot of horsepower: AI systems are generally divided into two parts: Training and inference. Training requires a large amount of specialized compute power, while inference thrives with the right type of device to provide a great user experience. Charlie Boyle, Senior Director of Product Marketing at NVIDIA, talked about the need for both of these parts to be robust in the AI pipeline, suggesting POWER9 servers and GPUs might be a solid option for training. On the inference side, he referenced voice assistants as an example and how these systems needed to provide a snappy a 7-millisecond response time to be considered acceptable for use by humans.
  • A strong holistic data strategy is essential for AI: AI systems require a significant amount of data to “train” or “teach” them to do a task or answer a question, so having a platform that can easily house a large amount of data and provide it to an AI system quickly and easily is extremely important. From the data perspective, enterprises need to account for the three main stages in an AI data pipeline: 1) Data curation/collection, 2) Training, 3) Inference. According to Vincent Hsu, Chief Technology Officer of IBM Storage, each of these stages require a certain focused characteristic of the data, ranging from easy access and programmability in the data curation stage, to high throughput in the training phase, to extremely low latencies in the inference stage. Although there are many ways to approach this, we’ve seen that an IBM data ocean strategy, anchored by IBM Spectrum Storage, can provide the framework for all stages.
  • The time for AI is now for enterprises, but not just because of the tech: In all the talk about how technology has finally evolved to accommodate AI at both a small scale for prototyping and a large scale for production, it’s easy to forget the human element and how the customers and stakeholders of enterprises and organizations benefit from using AI in their day-to-day lives. According to Dez Blanchfield of Gara Guru, one of the biggest reasons that AI is taking off today is that humans are finally understanding how to interact with AI systems and even “expect” them in everyday life. An example he gave is the automatic recognition from most people on how to interact and communicate with a voice assistant. There’s no need for AI if it doesn’t benefit customers, stakeholders, and the general population, so having people demand it is playing a key role in the current explosive uptrend of AI and shouldn’t be overlooked.
  • AI is a journey and it’s most important to just get started: Taking advantage of AI is extremely powerful and can seem like a daunting task, but we’ve seen it happens naturally over time through a highly iterative and collaborative process. Get started on a small, focused AI initiative with a reachable goal and see where it takes you!

Where can you start your journey?  Take a look at PowerAI!