Emma McGratan de Actian discusses the main challenges of today’s panorama and why the AI models are as good as the data that feed them.
Emma McGrattan is director of Technology (CTO) of the US Software Company. After obtaining an electronic engineering title from the University of the city of Dublin, McGratan initially joined Actian to work in the Line of Products Input in Dublin as an associated engineer.
“When I started, I made the promise that if I ever feared going to work, I found something new, and here I am 33 years later,” he says.
At the years, McGrattan has worked on a variety of functions in Actian, including the basic solutions of analysis, data integration and data management of the company.
In its current role as CTO, it leads the technological strategy, innovation and the development of products of the company.
“My goal is to position Actian to become a leader in data intelligence and ensure that customers, partners and analysts understand what makes us unique.”
What are some of the greatest challenges he faces in the current IT panorama and how do you go to them?
The data is very complicated and data ecosystems are very complex. Each organization we are talking about has data in multiple different types of databases and data stores for different use cases. As an industry, we must recognize the fact that no organization has a completely homogeneous data stack, so we need to admit and connect to a wide variety or data ecosystems, such as Databricks, Google and Amazon, regardless of the fortion, for the Dataalibst lineage and the like.
In Actian, we recognize the fact that they have this great disaster, and we want to help simplify it for them. We believe in decentralization of data with strong data governance, which allows the client to have their live data wherever it naturally adjusts to the insepasses of forcing them to build a warehouse. As a result, data integration and data lineage become important to understand the complete life cycle of the data, from compilation or creation to use, storage, preservation and final elimination.
At the same time, traditional centralized data governance models create bottlenecks that slow down innovation and decision making. We believe in federated governance, which combines centralized standards with specific domain flexibility. The person closest to the data decides who can access him for what use and purpose, etc., provided that he adheres to the centralized standards of organizations.
What do you think about digital transformation in a broad sense within your industry?
The adoption of the cloud is causing organizations to rethink their traditional data approach. Most use cloud data services to provide a shortcut to data integration without problems, efficient orchestration, accelerated data quality and effective governance. Actually, most organizations will need to adopt a hybrid approach to address all their data panorama, which generally covers a wide variety of sources that cover both the cloud and in the facilities.
What great technological trends do you think are changing the world and your industry specifically?
It is an exciting moment in the industry, since I see the impact of AI today as equivalent to when the world network was conventional 30 years ago: it is a complete paradigm shift.
I am more excited to give customers confidence to administer their AI initiatives on all the data on which your business is based. This is something that is the best for everyone in the world of data. Most of the AI projects that see implemented in production do not include the business crown jewels, which are the real data of the client. Companies are nervous about being wrong and the financial and reputation damage that could be captivated when making mistakes with the data or data of customers who are subject to regulatory requirements such as GDPR could be catastrophic.
‘Security has to be a central part of culture’
Another great challenge of adoption of AI is the process that requires a lot of time to build a semantic layer that understands everything about the business, including different domains, either finance or accounting or other department. If you are done in the traditional manual way, discover and label all data sources in an organization could be a project that never comes out. People do not have enough bandwidth or width, or in some cases patience, to ensure that everything is labeled as appropriate.
The application is a automatically generated rich semantic layer that allows direct conversations with the data. That is one of our objectives: allowing a commercial user to have a conversation with the data and is provided with rich responses that include visualizations such as graphics, graphics and trends.
As IA becomes more central to commercial operations, a truth is still clear: even the most sophisticated models are as good as the data that feeds them. I cannot emphasize enough the importance of the quality of the data, which I see in two ways. First, there are objective factors such as precision, integrity and punctuality. Equally important is the confidence and purpose of the data, which is much more subjective. Understanding the quality requirements dependent on the context of the thesis is crucial to prepare data for successful implements.
What do you think about how we can address the security challenges that are currently facing their industry?
Data safety has become a problem of trust, not just a technological problem. With AI, hybrid clouds and complex supply chains, the attack surface is massive. We need to design in mind from the safe coding of a single day, data level controls and zero confidence principles.
For AI, governance is critical, and must also be designed in a last moment idea. That means tracking where the dates come from, how models are trained and guarantee transparency and equity.
Security has to be a shared responsibility and a central part of the culture, not just a compliance box to mark.
Do not miss the knowledge you need to succeed. Register for him Daily letterSilicon Digest Republic of Science Fiction Technology News.
