Unprecedented confluence of tech mega trends
18 October 2019 | 0
The current range of technological forces at play represent an unprecedented situation globally, according to Patricia Florissi, vice-president and global CTO for sales, Dell Technologies.
Despite this, said Florissi, “We are technology optimists at Dell Technologies”.
Speaking to TechPro Magazine after her appearance at the Dell Technologies Forum in Dublin, Florissi said there are currently nine major technology forces at play.
These are artificial intelligence (AI) and machine learning (ML), the Internet of Things (IoT), augmented and virtual reality (AR/VR), robotics, specialised processors, the move from data centres (DC) to centres of data, blockchain, 5G, and quantum computing.
“These are the topics to which we must pay attention, and which our customers are going to use to drive differentiation in the market,” said Florissi.
She said customers are applying these technologies for core transformation in enterprise technology projects in the areas of application transformation, IT transformation, security transformation, and workforce transformation.
Florissi said that in AI there is something of a “perfect storm”.
Three enabling aspects are making AI more viable than ever, and that an interplay with the burgeoning world of IoT would only add to the effect.
Firstly, Florissi said, the complexity of the algorithms being developed, combined with the complexity of the models has enabled greater development than ever before, along with more specialised supports.
Secondly, the availability of data, combined with the systems’ ability to deal with it, means that it is easier to train model the complex models.
Finally, all of this requires levels of compute power and capacity that are now widely and cheaply available.
Florissi said that while the first concepts of AI came as early as the 1950s, and it has gone through many phases since. Only now though, has it come into its “spring”, due primarily to the conjunction of these three conditions.
Focusing specifically on the data aspect, Florissi said most people, when referring to data, are preoccupied with the size of data. “I say you have to very, very careful with what you wish for, because in AI, it is not just about the size,” she said.
By way of example, she says 10 million pictures of one person will not necessarily build a resilient model to do face recognition, because all the model will learn is to how to recognise that one face, if at all.
“In AI, what you need to address is what we define as data posture. This has three elements for the profile of data that you need for AI,” said Florissi.
“That can only come with IoT.”
Data posture, she said consists of three elements, dimensionality, duration and diversity.
Dimensionality refers to a level of depth and variation in the data, or many dimensions of the same scenario. The example is for autonomous cars, whereby multiple cameras can give different aspects of the same view that can be combined to build a better picture than any one or two views alone. This diversity of dimension can allow a model to be better trained to understand perceive and differences.
Duration in data means that events need to be captured longitudinally, over time, to provide a timeline. Again, this gives another plane of reference on which the model can organise.
Finally, diversity refers to diversity of scenarios to that allow the model to build resilience. A car that learns to drive in California is not necessarily going to be able to drive in Dublin without additional training for the specifics of that environment.
When asked about the preparation and cleansing of data to make it fit for model training, Florissi said that latest tools to develop AI are better than previous generations and can better handle such issues. Consequently, she said that there is less need for this discipline and believes it will only improve to the point where it may become a non-issue in the future.
However, looking at the mega trends, the 9 technologies aimed, Florissi said that the situation is unprecedented, but that two scenarios will play out.
Florissi argues that all nine are going to become interwoven in our daily lives and will cease to be a point in a couple of years. In much the same way that virtualisation has become woven into the fabric of data centres, enterprise and cloud architectures to the point of being invisible, so too will these nine headings.
Furthermore, she said that the combination of these technologies is what will lead to the disruption to come.
“I believe that the disruption is coming from the fact that those nine are happening all at once.”
“If I look back in computer science, I have never seen in my lifetime, nine things that could as disruptive happening at the same time, and think that defines this special moment that we are in.”
“One enables the other too,” she added. “One cannot have AI without IoT, one cannot have IoT without 5G. You cannot have 5G and all of the applications, and the continuum of the cloud without blockchain.”