Caragh O’Carroll spoke about three emerging technologies at Women in Tech Dublin 2018 : Blockchain, robotic process automation, and artificial intelligence and machine learning. She explored how these technologies provide solutions to the challenges that businesses are facing.
O’Carroll , a Fujitsu distinguished engineer, quoted the Gartner definition of blockchain: "A blockchain is an expanding list of cryptographically signed, irrevocable transactional records shared by all participants in a network." It’s a distributed technology with shared control, she said. Use cases for blockchain are for instance invoice checking, smart contracts, asset ownership, intellectual property tracking, audits, and compliance.
Blockchain helps you to build in trust, said O’Carroll. With blockchain technology, you can ensure that no information has been changed or tampered with. She suggested to apply them when there is a need for verification. O’Carroll suggested doing a proof of business to find out if blockchain can be valuable for a business, which is something that can be done in five days.
Robotic Process Automation (RPA) can be used to automate tasks previously done by human beings, said O’Carroll. It is often applied to repetitive and mundane tasks – the ones often seen as boring. With RPA you can have a robot doing it for you, she said. Solutions based on RPA technology have decisions built in which enable you to do creative work. She explained how you could train a robot to do purchase orders by building rules to extract information from an email, enter the information into the purchase order system, and generate the purchase order.
O’Carroll mentioned use cases for RPA: case management (for instance in healthcare), HR for administrating joiners, movers, people leaving, and banks. It can be cheaper to do these activities with robots, and automation can give people more time to spend with customers, she argued.
Machine learning (ML) and artificial intelligence (AI) are a different kind of technology as they are based on how our brain works with neural networks, said O’Carroll. It’s about predicting the right answer and getting better at it. You have to bring in data scientists, argued O’Carroll, as they know the techniques and how to apply them.
The main steps in applying artificial intelligence are defining the questions, gathering data, creating algorithms, and training the systems, said O’Carroll. It takes patience to apply AI and ML, she argued; you need to train systems, review the output, and refine to improve them.
InfoQ spoke with O’Carroll about blockchain, robotic process automation, and artificial intelligence and machine learning.
InfoQ: What is the status of blockchain technology when it comes to meeting regulatory and compliance demands?
Caragh O’Carroll: Blockchain gives us a great way to:
- Add accuracy - a blockchain ledger can provide a traceable record of an asset’s ownership, and ownership changes over time.
- Secure transaction detail – these ledgers are very difficult to alter because of their nature - multiple identical copies of the database are publicly shared (in a public blockchain). This also limits vulnerability to hacking, as an attack would have to target all the copies of the ledger concurrently.
For any areas with a regulatory or compliance impact, at the moment it’s still quite immature. The European Commission’s Action Plan on FinTech aims to support new business models and the adoption of new technologies while organising for EU-wide cyber risk testing, EU crowd funding and an EU public blockchain infrastructure. In parallel, the European Banking Association is actively analysing the impact on incumbent institutions about new technologies such as blockchain and also is assessing any consumer protection issues that arise e.g. in money laundering.
Locally, in Ireland, the Central Bank has put in place an Innovation Hub. The goal is that companies can share ideas and information to help ensure that future financial services regulation will be fit for purpose. There is no doubt that blockchain and distributed ledger technologies work have provided an answer to the key question in payments technologies – whether the person or organisation wanting to send the money, actually has it.
However, while blockchain and distributed ledger technologies are in theory transparent and tamper-proof, currently consensus on the standards and versions of blockchain is lacking. The technology is still evolving, and therefore my advice is to experiment but not move wholesale to blockchain until the clear standards and leaders emerge.
InfoQ: What are the benefits that robotic process automation can bring?
O’Carroll: Robotic Process Automation is the technology that allows you to configure a "software robot" to act as a human using IT systems as part of a business process. The robot uses the same screens (user interfaces) as a human and can capture data and manipulate applications in the same way. Just as we may look for certain data in order to trigger certain activities (e.g. an email with an order or query needs data put into another email or a system to move the process on), a robot can learn this activity. The goal is to use the robot for repetitive tasks.
The key benefits are:
- Zero typo or copy/paste mistakes
- Can work 24x7 if you want (but watch the impact on downstream processes)
- Time saving – spend time on more valuable activities e.g. new business development
- Cost saving – possible rationalisation depending on volumes
- Compliance – by always doing the correct thing, compliance is a great byproduct
Due to the impact on job roles, RPA does require a change programme to ensure that time freed-up is focused where the business needs it most, and that those benefiting from or impacted by RPA are involved in the change and realising the benefits.
InfoQ: Why would we embed AI and ML into business applications, what are the benefits?
O’Carroll: There are a range of benefits and they all start with the question(s) you want to answer. However, answering that question isn’t so easy – AI and ML are seen now in digital assistants (think Siri, Alexa), facial recognition (e.g. Sensory Inc.), photo captioning (tagging people) and product recommendations (targeted ads).
The benefits though can be considerable:
- Using AI/ML in the DevOps world to streamline the release process – and move to a more Agile CI/CD approach. Use this to iterate testing, tuning, and releases.
- Building in business Key Performance Indicators not only to report, but also for scenario planning e.g. the impact on the system or process for changing a KPI.
- Capture, analyse, personalise, recommend re: customer data – based on multiple input streams, would a different price plan or an upsell of an extra product be recommended?
While business applications today rely on access to an external database (Oracle, SQL) or in-memory database (NoSQL), future applications will include APIs for AI, e.g. converting speech to text in contact centre recording to identify customers or callers with certain sentiment, e.g. anger or likely to leave (churn) in near real-time. This type of solution could also help with live translation for those servicing e.g. patients who are not fluent in English… the text can also be converted to the spoken word in the target language.
Another option is to create an in-house machine learning platform based on bringing together data from multiple internal and external sources. Once the right questions are identified, the appropriate algorithms need to be selected and then the model trained before moving to production.
InfoQ is covering Women in Tech Dublin 2018 with Q&As and summaries. Earlier InfoQ published The Importance of Feedback for Skill Development and Careers and Embracing Diversity and Fostering Inclusion: A Necessity .