For all the hype and buzz around it, the actual adoption of artificial intelligence (AI) has not really taken off in the enterprise domain yet. A Forrester report at the start of this year revealed that a lowly 12% of the included tech executives had started using AI tools and systems full-fledgedly. However, the immense opportunities of AI solutions are not lost on business entrepreneurs – with 6 out of every 10 enterprises (in a study involving 235 business professionals) stating that they will start using AI technologies by the end of 2018. The mounting interest in AI is also reflected by the fact that revenues from enterprise applications powered by the technology will jump to more than $31 billion in 2025 – growing at a CAGR of ~63% (in 2016, the corresponding revenue figure was $359 million). In today’s discussion, we take a look at how artificial intelligence is shaping enterprise applications and paving the way for valuable innovations:
Increasing awareness and greater mindshare
In a recent PwC survey in the United States, around 73% prominent business leaders opined that artificial intelligence will be the biggest game-changer – something that delivers significant competitive advantage – in the foreseeable future. Nearly 40% of the respondents are looking out for suitable identification and designing solutions for enterprise AI deployment – while a large group of businesses are currently carrying out various types of AI-related researches. In a separate study (with 146 respondents), it was found that ~80% entrepreneurs are prepared to make significant investments on AI technologies in 2018 (a large number of ‘Proofs-of-Concept’ are being evaluated; many of them are already live). A measly 3% of all enterprises are not yet aware of the concept of artificial intelligence – although there are several companies (22%) which do not have the budget/financial means to integrate AI systems in business IT.
AI technologies in action
While most discussions revolve around how AI is revolutionizing the world – it is not a technology per se. Instead, think of it as an ‘umbrella layer’ or a ‘horizontal enabling layer’, which contains a large number of cutting-edge technologies. These include machine learning and deep learning, natural language processing (NLP) and natural language generation, decision management systems, biometrics tools, text analytics, robotics process automation, software and hardware platforms (for both ML and DL), and other related tools. When an enterprise opts for AI, it does not purchase or implement the ‘entire package’. Instead, it looks for the technology(ies) that would deliver the maximum value for business (ROI). Virtual agents (referred to as ‘the current darling of the media’ by Forrester) are yet another popular example of AI-tools that have made a difference (think Siri, or Alexa, or Cortana). AI-powered chatbots are also witnessing healthy adoption rates. All the technologies are, in their own way, geared to automate processes, increase operational speeds, and deliver optimal results.
Domains of enterprise transformed by AI
By the end of this year, the value of the global enterprise market will touch the $3.5 trillion mark. Artificial intelligence and machine learning are affecting an ever-increasing share of the overall market – and interestingly, nearly all the important sub-domains of enterprise are being transformed by the technology. Right from security & risk handling, business intelligence, knowledge processing, productivity management and data science management, to B2B digital marketing/sales, e-commerce, finance operations, manufacturing, customer management, and even drones and (maybe) robots – AI is making its presence felt practically everywhere. This growth is being fueled by the fact that the number of AI vendors/AI companies are rising rapidly across the globe. At last count, the total number of AI startups had crossed the 2250 mark – and remarkably, almost half of these startups have been launched in the last 2-3 years.
Moving beyond machine learning
Data is the backbone of all artificial intelligence technologies. The so-called ‘AI winter’ – during which excitements over AI were seemingly slowing down – was primarily caused by the fact that most enterprises did not know what to do with the accumulated data (for many businesses, labeling the available unstructured data was also a considerable challenge). As machine learning technologies evolved, data could finally be used for predictive purposes. However, as we step into the next phase of AI development for enterprises, ML is not going to be sufficient in itself – and efficient ‘machine intelligence’ (MI) will become the more important capability. While ‘machine learning’ can estimate/predict that ‘X’ units will be produced this quarter or ‘Y’ number of shipments will be made in a year – ‘machine intelligence’ goes a step further, and delivers deeper insight into the ‘why’ factor – the causes behind these estimates, and whether (& how) such predictions can be modified. The focus is firmly on doing more with big data – and machine intelligence has big capabilities in this context.
Where are the AI investments being made?
As we enter into the ‘golden age of enterprise AI applications’ – with mounting investments by businesses in a bid to gain greater tractions in terms of quality, productivity, scalability, and efficiency – ‘machine learning applications’ have a clear lead over other technologies (with $3.5 billion funding). The next three spots are occupied by ‘natural language processing’, ‘smart robots’, and ‘machine learning platforms’ – with ~$1 billion investments on them respectively. Tools like ‘image recognition platforms/applications’, ‘speech recognition’ and ‘virtual agents’ also attract healthy investments from enterprises. Slightly surprisingly though, the least interest (in terms of investments) is on ‘video content recognition’ and ‘speech to speech translation’. Over the next half a decade or so, these technologies are also expected to push upwards.
The importance of establishing data-driven culture in enterprises
The technical capabilities available – high-end ML and DL platforms, advanced algorithms and seriously high computing power (by 2020, we are looking at ubiquitous workplace robots and ‘speed-of-light’ computing power) – might be top-notch, but enterprises that do not have a data-driven culture are not going to make much headway in the AI space. Any company that is serious about implementing artificial intelligence in its existing IT infrastructure simply must have a proper team of chief data officers (CDOs), data scientists, domain experts, information engineers, statisticians and other such experienced professionals – who can i) understand and effectively analyze the huge volumes of data, and ii) collaborate with customers, decision-makers and fellow employees properly. These ‘data professionals’ will have to work with enterprise app developers, in order to chalk out plans of integrating AI and ‘intelligent’ data flow in new applications. The onus of testing software tools and frameworks, along with the capabilities of different AI services, will also be on these ‘data experts’. The recruitment, training and talent-sourcing processes in enterprises have to be suitably modified.
APIs for internal AI projects of businesses
There are many use-cases of high-end application program interfaces (APIs) being used for streamlining in-house business AI projects. A classic example regarding this would be Ocado – which has successfully brought together robust cloud APis with Google TensorFlow – in a bid to manage the high daily volumes of customer emails received by the company. There is also considerable buzz over how the company has plans to ditch traditional packing procedures with barcode scanning, in favour of ‘intelligent AI vision’ practices. ML algorithms are being implemented within both end-user apps (e.g., Google Home) as well as other enterprise tools (e.g., routing applications). APIs can be used to make apps ‘smarter’ in a myriad of ways – like integration of NLP, or video search, or text-to-speech/speech-to-text capabilities. The scopes for switching over to AI-based tools and solutions is uniformly high across the board, and more innovative deployments should become active in 2018.
The benefits of AI applications
Given the potentially substantial advantages of AI technologies in the enterprise domain, it can be safely said that these will continue to grow in popularity over the next few years. Applications with built-in artificial intelligence can give current enterprise IT setups a total makeover – with agile software development and release, ‘intelligent’ data diagnostics, rapid and accurate processing standards, process dissipations, and holistic digital experiences for users. There are considerable benefits from the financial perspective too. By bolstering the overall efficiency of operational flows and bringing down the cost of manpower maintenance/management, AI will be able to help US-based businesses save up to $60 billion. AI tech is finding acceptance in different industrial sectors – led by finance and insurance (understandably, with financial data management and fraud detection being key activities), along with medical and healthcare, education, manufacturing and transportation. Predictive maintenance systems have made it possible for businesses to follow ‘just-in-time (JIT) maintenance’ standards. Inventory levels, hence, can be trimmed – and that has been a big advantage as well.
AI use cases in enterprise software
The importance of AI solutions is healthcare is underlined by the fact that the technology yields a massive $2250 million from customized processing of patient data, with built-in scalability. However, it is far from being the most common AI use case in business software applications. The top spot in this regard would go to algorithm-based improvement of trading strategy performance use cases (revenue $2400) – with static image recognition and tagging occupying the second spot. Other common, and significantly revenue-yielding use cases of AI-powered enterprise software applications include social media content distribution, text queries of images, object detection/identification/classification, and predictive maintenance. As organizations ‘learn’ to optimally deploy AI technologies, more and more new use cases are sprouting, and revenues are skyrocketing.
The human-displacement consideration
It has widely been speculated that large-scale implementation of automated solutions (say, chatbots) will result in huge loss of employment worldwide. This is, truth be said, only partially true – since the human element is still going to be required to control all the gadgets and algorithms and software (even when machines are ‘intelligent’ enough to not require any prior programming). As highlighted in an earlier, without the ultimate control in the hands of capable human data experts/scientists, the technology itself is hardly going to be of any use. There will be some labor-displacement though – thanks to focus of AI on making all menial, repetitive human activities automated and smarter. As AI becomes mainstream in the next 4-6 years, low-level employees might find themselves out of work (‘replaced’ by the considerably more efficient machines). AI is not going to affect those working at the top rungs of enterprise IT and management systems. Unless we are talking about a dystopian future, AI machines and robots are not going to order human beings around anytime soon!
The steady rise of cognitive computing
The spurt in the growth of AI solutions for enterprise is boosting another form of innovation – in the form of cognitive computing (through a combination of AI and signal processing capabilities). These fully autonomous computing systems do not require pre-programming, and have – in varying degrees – 4 key abilities – ‘to sense’ (through IoT ‘sense-and-respond’ networks), ‘to learn’ (by drawing ‘informed’ conclusions by making use of past data/experiences), ‘to infer’ (by working just like the human brain, with the help of powerful AI algorithms), and ‘to interact’ (via custom natural language interfaces through gesture, voice or touch). By the end of 2020, cognitive computing will be a $13.7 billion market globally – and it has tremendous potentials of taking enterprise operations and applications to an altogether higher level.
Arrival of Intelligent App Stacks
In less than five years from now, 4 out of every 10 enterprise apps will have AI and ML capabilities. Adoption of ‘intelligent’ technologies are helping enterprises seamlessly move over from the traditional ‘system-of-record’ infrastructure, to a much more efficient ‘system-of-intelligence’ network – with apps, APIs, platforms and priority systems. Such ‘Intelligent App Stacks’ will have favourable effects on the entire value chain of organizations – pulling up overall ROI figures and limiting unnecessary expenses. An ‘intelligent’ enterprise app can easily influence customer behaviour, push out real-time updates, generate personalized responses, and perform a host of other high-end functions. Over the next few years, both infrastructure-as-a-service (IaaS) and machine learning-as-a-service (MLaaS) are going to become increasingly mainstream.
Given the range of opportunities and capabilities that enterprise AI opens up, the mounting interest of the biggest tech players in the technology does not come as a surprise. From IBM (with Watson) and Google (with the $400 million acquisition of DeepMind), to Microsoft (with its venture capital startup for AI), Facebook (AI-powered photo narration on the iOS app) and Uber (with last year’s acquisition of Geometric Intelligence) – all the market leaders are investing on artificial intelligence in a big way. We are still a long way off from making robots or machines that surpass (or even match) human intelligence on their own – but AI-powered solutions are certainly poised to change enterprise software as we know it.
Latest posts by Hussain Fakhruddin (see all)
- AI-as-a-Service: All That You Need To Know - October 30, 2018
- Top 12 Trends In Digital Transformation To Watch Out For In 2019 - October 23, 2018
- Machine Learning in 2019: Tracing The Artificial Intelligence Growth Path - October 9, 2018