From the March 2018 Issue.
Emerging technologies have always changed the practice of accounting. Further, your clients’ businesses also change based on their choices to use technology. Emerging technology is no long “emerging” when it becomes mainstream. For example, Optical Character Recognition (OCR) has been evolving from an emerging technology to a mainstream technology. If you use an expense reporting product like Zoho Expense, a 1040 workpaper product like CCH SCan, or a note taking tool like Evernote that does OCR, you are beginning to appreciate how well OCR works now compared to a few years ago.
As the second column in a lengthy series on emerging technologies, it is time to begin building a framework to understand emerging technologies. Most of these emerging technologies have a relationship, much like the Grand Unified Theory (GUT) of particle physics. You’ll see the overall relationships in a later column. In the first column, you were introduced a structure to understand each of the emerging technologies in a simple table. Each why and how column will contain a table summarizing the main concepts, as you can see below. Now, for the subject at hand.
Cognitive Computing (CC) is based on the scientific disciplines of artificial intelligence and signal processing. These platforms encompass machine learning, reasoning, natural language processing, speech recognition and vision (object recognition), human–computer interaction, dialog and narrative generation, among other technologies. Cognitive Computing applications link data analysis and adaptive page displays (AUI) to adjust content for a particular type of audience. Another definition of CC is: hardware and/or software that mimics the functioning of the human brain and helps to improve human decision-making.
Because of the way researchers position CC, it is close to the GUT of emerging technologies. Making computers act and respond like humans is what CC is all about. Various research areas of computing largely lead back to CC.
Why?
After reflection, you’ll find that the concept behind Cognitive Computing is obvious. The goal is to build sufficient technology to meet or exceed what a human can do in the same role. While this sounds futuristic, much progress has been made in the last 40 years, particularly after a major shift in the strategies, called algorithms changed in the 1990’s. Instead of trying to program computers to do each step, computer scientists and developers started using statistical models to help computers learn. These statistical machine learning techniques are used by companies like Receipt Bank and Zoho in the software products they have brought to market.
But note that the definition of CC includes data analysis and adaptive page displays. Applications that apply CC methodologies learn, analyze and present the data in a clear, usable way faster and more accurately than can be done by a person. The AUI approach is supported in browsers and mobile devices by HTML5 and responsive web design. Receipt Bank 1Tap is an example of how these technologies are being used today. While the goal is 100% accuracy in all computing, algorithms are not perfect. But then again, neither are people.
However, 1Tap is the closest piece of technology to 100% accuracy in the field of accounting today. Applications will continue to use one-upmanship just like we saw in the early days of spreadsheets with Lotus 1-2-3, SuperCalc and Excel each introducing new features and innovation in each release. The pace is frenetic, although sometimes not well-tested in SaaS products today because managers don’t understand how to properly manage developers in scrum techniques…but that’s a topic for a different column. Competition was good and led to superior functionality.
So why do we want cognitive computing? To provide automation of routine tasks giving us and our clients time to focus on the more important business goals. As I learned from Brock Philp, now CEO of Newforma in 2016 while he was President of Doc.It workflow and document management systems, “If you say it real fast, it sounds easy”, applies to cognitive computing and most emerging technologies. CC has required a mighty effort to build and understand, and the hardware and software computing resources needed to make it all work well have just become available in the last ten years or so.
How?
So how do Cognitive Computing approaches work? They are:
- Adaptive: They may learn as information changes, and as goals and requirements evolve. They may resolve ambiguity and tolerate unpredictability. They may be engineered to feed on dynamic data in real time, or near real time.
- Interactive: They may interact easily with users so that those users can define their needs comfortably. They may also interact with other processors, devices, and Cloud services, as well as with people.
- Iterative and stateful: They may aid in defining a problem by asking questions or finding additional source input if a problem statement is ambiguous or incomplete. They may “remember” previous interactions in a process and return information that is suitable for the specific application at that point in time.
- Contextual: They may understand, identify, and extract contextual elements such as meaning, syntax, time, location, appropriate domain, regulations, user’s profile, process, task and goal. They may draw on multiple sources of information, including both structured and unstructured digital information, as well as sensory inputs (visual, gestural, auditory, or sensor-provided).
What does this mean to the practice of accounting and to accountants? We have several working examples available from larger companies:
- IBM Watson – Tax processing for New York State, summarized on an innovative web site for marketing by the firm R&G Brenner Income Tax
- KPMG – Audit Services with IBM Watson, reported by AI Business
- Deloitte – Center for Technology, Media and Telecommunications
The next step will be to get these approaches simplified, less expensive and gain the ability to work in the mass market with mid-size and small accounting firms and their clients.
As development continues and cognitive computing transitions from an emerging technology to a mainstream technology, they will choose from many open source and proprietary suites that have cognitive computing capabilities. Examples today include:
- SparkCognition
- Microsoft Cognitive Services
- Expert System
- IBM Watson
- Numenta
- Cisco Cognitive Threat Analysis
- Facebook Deepmind
- Customer Matrix
- Cognitive Scale
- HPE Haven OnDemand
In future articles, if there are other examples of products available today that are working, they will be included here. We are not convinced that many of the vendors really have cognitive computing applications working. They are simply riding the band wagon of popular marketing terms or the latest fad.
For example, at CES 2018, it was clear that there were emerging technologies that would become strategically important as brilliantly summarized by Richard Quinell, Editor in Chief of Electronic Products in his article, “Peek behind the curtain at CES 2018 reveals innovations to come”. However, Brian Tankersley and I concluded after our week at CES 2018 that there was a lot of “artificial, artificial intelligence”. Just like the first column in this series referred to a lot of “want to be” players in emerging technologies where their claims are ludicrous, loony, and ill-informed, there are a lot of band wagon/want to be vendors who frankly don’t have the technology or tools available or in use now.
When you see a tool listed at this point in future articles, you’ll know that it has been vetted to be the “real deal”. The best example of a tool for accounting that is working today is:
- Receipt Bank 1Tap, used for small business tax Schedule C automatic accumulation and classification. With the company’s current strategy, you can pay one annual fee and use this product with an unlimited number of clients. It is a marvelous value.
Here’s a summary of what you need to know about Cognitive Computing:
|
Key Information |
Technology: Cognitive Computing |
|
Why is the new technology better? |
It is the overall container for most emerging technologies |
|
How can you do this today? |
|
|
Risks |
Need to properly define needs |
|
Where/When to use |
When you need human/machine interfaces |
|
How Much? |
Depends on platform |
|
Displaced technology or service |
Traditional computing |
|
Other resources |
Cognitive Computing is among the most difficult of the emerging technology concepts to grasp because to understand it fully, you need extensive computer science background. However, I’m reminded of the saying “don’t tell me how to make a watch, when all I want to know is the time” whenever emerging technology is discussed.
As a practitioner of accounting, you have real world client problems to solve, and just want to know which solutions work, what they do, and which products or approaches you should avoid. That will continue to be my job in future Why and How columns.
Recommended Next Steps
Watch for applications that have automatic classification and can interface to many different systems. Products that are just arriving in the market will probably be built with new generation development tools that leverage emerging technologies. There will be hype and lies, so you’ll need to be careful that you don’t get a solution that is not built properly to work at scale and in all situations. We saw that mistake made over the last ten years with 1040 Workpaper products. With today’s development tools (SDKs and APIs), it is possible to build a more sophisticated product rapidly. One caution that even the developers frequently miss: be aware of vendor lock-in. Vendors want to tie you to their products, much like the Eagles’ old song “Hotel California”. “You can check out any time you like, But you can never leave!”. Be flexible enough and wise in your choices so you can chart your own destiny.
Thanks for reading CPA Practice Advisor!
Subscribe Already registered? Log In
Need more information? Read the FAQs
