The Semaphore - Smartlogic’s Quarterly Newsletter - 21Q4

CEO Corner

Explainable and Trustworthy AI – How does the Business Trust an Algorithm

Today’s AI applications and the concept of Machine Learning create a false expectation that we’ve reached The Singularity - when machines act as fully functioning autonomous units that perceive, learn, decide, and act on their own. In simple terms, Explainable and Trustworthy AI identifies AI that is transparent and reliable in its operations so that human users can understand and trust its business decisions.

Today, AI does not live up to the trustworthy/explainable task as it:

• Uses complex mathematical algorithms over vast amounts of data
• Opaque – the algorithms are applied by a specialist data science expert, not a Subject Matter Expert
• Unbelievable – the results are difficult to explain, understand, and often noisy
• Brittle – can fail over time due to unintended bias and is difficult to debug

The inability to explain doesn’t hurt us until we use AI for things that matter and, depending upon the seriousness of the application and consequences of error, low-level explainability might be enough. In more serious applications, such as compliance tasks in Financial Services and clinical diagnosis in healthcare, a very high degree of explainability is required. When not provided, businesses could face large fines or damage their reputation.

If you understand its limitations and don’t expect perfection (we don’t with our human workers after all) AI can be transformational for businesses. It can increase work efficiency, work with high accuracy, reduce the cost of training and operations, and improve processes. However, evidence indicates that AI works best as an addition to human skills, not as a replacement for them.

AI systems must be:

• Robust – It must operate consistently over time, demonstrate persistent levels of accuracy in its reasoning, and operate in context to the data.
• Trustworthy – Results are believable, auditable, deterministic, and explainable to business users.
• Fair – Treat past and current data consistently so that it does not contain unintended bias or discrimination; keep track of how data is collected, where it is from, when it was collected, and how data is used to determine any result.
• Reliable - Prevent unwanted data translations or manipulations due to unintended or uncontrolled algorithm creep.
• Auditable – all information and algorithms related to the task should be governed, reported, and tracked by business users.

AI needs to be implemented with an understanding of the data context and to be able to refine judgments/decisions when new information is presented. Without context, AI is narrowly focused, makes subpar predictions, and has limited transparency. AI that is powered by semantics – Semantic AI – provides context to the human: machine interaction.

Semantic AI combines several AI technologies including Machine Learning, Natural Language Processing (NLP), and semantic reasoning as a collaborative interplay between humans and machines. This approach allows machines, as well as people, to understand, share, and draw conclusions from data - structured, semi-structured, or unstructured, internal, or external to an organization.

Semantics, i.e. the communication of meaning is a key component in AI as the understanding of meaning transcends industry, organization, and use case. Semantic AI (the application of meaning on information using machines) enables safe and responsible applications and systems that are trustworthy, transparent, and align with societal norms and values.

Data in context, delivered from Semantic AI systems, can facilitate responsible AI. Metadata, (which is the output of Semantic AI) is used to reason, harmonize, and enrich data leading to robust and trustworthy results. Semantic AI addresses the issues of fairness and reliability by eliminating bias and providing explainable and transparent outcomes without the need for complex training sets.

In this way, Semaphore, our Semantic AI platform, removes the layers of information complexity and allows you to use enterprise data not just for knowledge but with the power of wisdom to act faster.

Semaphore:

• Deciphers the context of your data – all you need to know, starting with your data.
• Filters apply logic to identify, extract, and apply meaning to hidden information – giving you context.
• Connects enterprise information and eliminates the effects of information silos.

Semaphore integrates with content, data, and application systems; supports multiple data types, and uses AI and ML to transform digital data into qualified actionable intelligence so you make wiser decisions and reduce risk within your organization.

As we go forward, human values will increasingly impact AI systems and context will be a necessary component to make AI responsible, reliable, and safe. To move from the questions, we have today into what is required for future AI applications, Semantic AI is the necessary governing component.

Best Regards,

Jeremy Bentley

CEO & Founder

Semantic Enabled Data Fabric Powers Trade Management Application

Whether you operate in capital markets, asset management, consumer finance, or insurance, you need solid information to make decisions that drive positive organization, stakeholder, and customer outcomes. As the volume of data flowing into financial organizations continues to explode, the ability to rapidly access, manage, and make sense of it is key to managing risk and improving ROI.

Today, financial services organizations need innovative solutions that enable them to:

• Unify and harmonize a variety of data types from disparate sources to provide a single view of the organization.
• Provide relevant information to a broad range of internal and external stakeholders in real-time for analysis, reporting, and management.
• Ensure compliance with regulatory mandates to decrease risk and avoid sanctions.

The key to managing enterprise data is harmonization, the ability to provide a holistic view of all information, structured and unstructured, regardless of location and type that can be used to manage the business. Download this complimentary case study to learn how one organization powered their Trade Management Application with a Semantic Data Fabric.

Announcements

Provide an Info-Tech and/or Gartner Review for Smartlogic

As many of you know, Smartlogic is reliant on product evaluations from external resources like Gartner, IDC, Forrester, etc. that publish reviews, capability reports, and other documentation. These reviews help customers like yourself make informed purchases when evaluating technology products.

If you have already provided reviews for us on Gartner’s Peer Insights we’re very thankful for that. If you’re interested in providing a Smartlogic review to Gartner and/or Info-Tech – use the links below.

Info-Tech is an advisory service that is planning to publish a series of reports around Metadata Management Solutions. To be included in that report, Info-Tech requires that vendors have at least 10 customer reviews of their product. They are offering a $25 Amazon gift card (or charitable contribution) for customer reviews of our Semaphore platform, which you can access here: Review Smartlogic & Receive a $25 Reward.

Gartner – offers a 3-month limited subscription for completed and approved reviews: https://it.gtnr.io/GulrB7qUS

All reviews are anonymous – we won’t know who reviewed us. The time to complete a review is approximately 10 minutes. Your participation is greatly appreciated.

Events

Smartlogic’s User Community Forum

Our next User Community Forum meeting will be held on Wednesday, April 7th at 11 am Eastern. If you haven’t signed up yet DO IT TODAY – it’s a great opportunity to meet and connect with like-minded Semaphore users to exchange tips, tricks, and best practices with the Semaphore platform and Knowledge Management topics. We look forward to having you join us.

Press Releases

Smartlogic Retains Gold on Microsoft Competencies

Smartlogic, the leader in Semantic AI platforms, today announced it has retained Microsoft Gold Partner Competencies in Application Development and Cloud Platform added the Application Integration competency, and gained approval for their Semaphore Cloud platform in Azure.gov. Read the full release.

Smartlogic Releases Semaphore 5.2.0 – The Next Generation of Semantic AI

Smartlogic, the leader in Semantic AI technology, today announced the release of Semaphore 5.2.0, which includes new features and improvements across multiple Semaphore modules. Read the full release.

Training

2021 Spring Training

Semaphore Foundation Training is underway but there is still time to register for our Spring Advanced, FACTS, and Administration training. For registration and start dates, go to our events page.

Not sure where to start your training, visit our website training page to view pre-requisites and course information. To register for training, email training@smartlogic.com or contact your Smartlogic Account Manager.

Webinars/Recordings

If you missed our webinar - The State of Semantic AI with Jeremy Bentley on January 21st, 2021 you can use this link to listen to the replay.

If you missed our webinar - Semantic Data Catalogs – A new Route to Value with Steve Ingram and John Lukasiewicz on February 25^th, 2021, use this link to listen to the replay.

Coming Soon

More webinars are in the works, keep an eye on our events page and read our weekly Life with Semaphore for announcements and registration information!

Smartlogic’s Social Media

To learn about upcoming events, webinars, and recent blog posts, follow us on:

Website: www.smartlogic.com

LinkedIn: https://www.linkedin.com/company/smartlogic/

Twitter: https://twitter.com/Smartlogic_com

Facebook: https://www.facebook.com/SemaphoreSmartlogic/