How digital organisations can go “hybrid” in AI adoption
What is Hybrid AI and how does it raise the bar for digital organisations dependent on conventional AI methods? Learn how this disruptive technology can drastically change the way people work and engage customers.
Emerging technologies are now a thing of the past. With pressure to exceed expectations of tech-savvy citizens or customers primed by innovation in our post-digital world, organisations are beginning to look towards disruptive technologies that promise incredible results. This means technologies like social, artificial intelligence (AI), Cloud, Robotics, Touchless Transaction, and more. But which solutions can actually help organisations stay ahead of the curve?
A blend of statistical and symbolic AI makes “hybrid”
Newly contending the disruptive technologies landscape is a form of AI that merges the main two classes of AI—statistical and symbolic AI. Referred to as “hybrid AI”, the unique technology is part of the move towards broader AI systems that integrate multiple AI techniques. Such parallel combinations aim to overcome the weaknesses of specific systems that employ singular AI methods, especially pure statistical AI. One mainstream example of this is the heavy reliance on Machine Learning (ML), so much so that enterprises have often likened AI to just ML.
Can hybrid mitigate salient challenges faced by organisations?
But first, how does ML work? This methodology examines large volumes of data—which must often be uniform in nature—to derive statistically probable correlations. ML performs relatively well in finding patterns in uniform data. Realistically, however, situations with such ideal data consistency and volume are few and rare.
In fact, a webinar poll conducted among public sector employees showed overwhelming resonance for this concern. Results showed that the highest ranked barrier faced in adopting AI was the “Lack of suitable data volume or quality to apply AI solutions”. Respondents included public officers from Singaporean government agencies who attended the jointly organised virtual event, TAIGER x IMDA Digital Concept Series: Hybrid AI in a Digital Government.
Poll results collected among public officers in the event, TAIGER x IMDA Digital Concept Series: Hybrid AI in a Digital Government.
The world we live in today is predominantly unstructured—along with the data we deal with. Analysts from IDC to Gartner have concurred with an estimate that around 80% of data worldwide is already, or will soon be, unstructured. “Nearly 80% of data in the enterprise is unstructured—work descriptions, résumés, emails, text documents, research and legal reports, voice recordings, videos, images, and social media posts,” according to a 2020 sharing by Accenture. Organisations jumping on the ML bandwagon would have unfortunately realised that ML systems experience tremendous difficulty in understanding non-standardised or unstructured content.
Logic driven hybrid AI better mimics the human brain
On the other hand, symbolic AI is driven by logic. This logic or rule is provided by humans and pre-taught to symbolic AI systems, which enables them to understand minute data sets using the given rule. When symbolic is employed in parallel with non-symbolic AI, it closely resembles natural intelligent systems, i.e. our human brain. The hybridisation similarly possesses common sense and knowledge about concepts and their relationships. This means that hybrid AI can solve operations that need a lot of thinking, reading, reasoning and deduction.
This new disruptive technology is turning the heads of industry honchos such as Gartner. In fact, there seems to be a growing common understanding that “the hybrid AI approach is the more complete approach to natural-language automation”, said Anthony Mullen, a senior director analyst at Gartner.
Cognitive tasks are where hybrid can best be put to work
Hybrid AI’s ability to think like a human makes the technology highly versatile for organisations in tackling complex real-world problems. Examples include extracting meeting minutes from text-heavy emails, or even teaching a chatbot to understand slang and respond accordingly. Such tasks are far beyond certain emerging technologies like Robotic Process Automation (RPA) which has seen popularity among tech adopters, albeit in automating less cognitive operations.
Where does TAIGER sit in the hybrid AI space?
For Singapore-based AI startup TAIGER, hybrid AI is realised in an automation platform Omnitive which brings together a solution suite of automation tools: information extraction, semantic search, virtual assistants, and no-code application development. These tools integrate multiple branches of AI—including ML, Natural Language Processing (NLP), Knowledge Representation and Reasoning—to give hybrid AI.
Non-symbolic and symbolic AI that make TAIGER’s hybrid AI
Navigating a starting point becomes the next concern. But thankfully, the technology’s merits have been proven in a broad range of industries from finance to retail and public sector to education. TAIGER’s proprietary AI models have delivered successful onboarding solutions with finance heavyweights like Banco Santander and Otkritie Bank. Even the most abstract forms of data such as arts and culture can be accurately processed with TAIGER’s hybrid AI approach. Lauded as AI that has the ability to read anything, the technology’s application is perhaps up to one’s imagination.
Explore more hybrid AI use cases and success stories for the public sector shared during the TAIGER x IMDA Digital Concept Series: Hybrid AI in a Digital Government.