Artificial Intelligence (AI) describes the ability of a machine to perform cognitive functions associated with human capabilities: specifically, learning, reasoning and problem-solving.
The AI label is sometimes loosely and wrongly assigned to the simplest of process automation tasks. However, a basic, task-specific automation system can only ever work by following set rules assigned to it – and these usually take the form of ‘if’ statements (‘if this, then that’). There is no understanding (i.e. intelligence) behind the action: the machine simply does as it is told.
By contrast, a machine with true AI capabilities is example-driven rather than rules-based. It can learn from the data it is exposed to: a process referred to as machine learning. It can also identify patterns and even make decisions with little or no human intervention.
AI involves ‘teaching’ a machine to make predictions and solve problems based on given parameters. With most forms of AI, two things are needed for it to work: powerful processing capabilities and large volumes of data.
Usually, the more data the machine processes, the better it becomes at interpreting it, at making predictions and solving problems.
Fraud prevention provides a useful illustration. A financial institution demands the ability to spot and block fraudulent transactions. Basic monitoring capabilities enable it to flag up specific events such as previously unseen log-ins and unusually large fund withdrawals. But once AI is added to the mix, this institution’s fraud prevention capabilities are significantly enhanced.
By processing user behaviour data, the AI-driven solution is able to recognise the ‘normal’ conduct of account users and detect deviations from these normal patterns. In turn, through a combination of machine learning and statistical analysis, it can identify which of these anomalies are at risk of posing a fraud threat.
In this way, the institution has a means of identifying potentially risky outlier transactions and behaviours – including previously unseen fraud methods. The greater the volume of data the machine processes and the longer it is in operation, the better it becomes at accurately assessing risk – thereby reducing the likelihood of false alarms.
For forward-thinking organisations, AI is the new normal.
Ask your device a question and instantly get the answer. Tell Alexa to complete a task - and she knows exactly what to do. Just a few years ago, this type of technology had a certain novelty value. Now, using it comes as second nature. Underpinning it all is the rapid advancement of Artificial Intelligence (AI).
The same technology is also powering a quiet revolution in organisations across the globe. Businesses are using AI to automate tasks and free up time and resources. They can unearth hidden insights from vast quantities of data. They cast new light on customer behaviour and are better able to anticipate customer needs. They can ask ‘What if’ and get answers they can trust.
These capabilities are now mainstream. It means that to stay ahead of the competition and to solve genuine business problems, organisations need to think about where and how AI might be usefully deployed.
Using intelligent algorithms (sets of rules), systems powered by machine learning have the ability to learn automatically through the processing of data. In the fraud prevention scenario above for instance, the solution is able to actively identify risky anomalies based on detailed analysis of behavioural data, rather than merely spotting red flags according to a specific list.
Think of this as a supercharged variant of machine learning. It relies on artificial neural networks – in other words, algorithms inspired by the human brain, to learn from large quantities of data. Deep learning enables machines to make sense of data inputs that are diverse and unstructured in nature (auditory questions being one example). Practical uses include virtual assistants, facial recognition and driverless vehicles.
Powered by machine learning, predictive analytics involves the analysis of historical data to highlight trends and patterns and identify the likelihood of future outcomes. For forward-thinking organisations, this subset of AI-powered technology is revolutionising decision making.
Augmented analytics (aka smart data discovery) is an advanced form of business analytics incorporating machine learning and natural-language generation. This powerful combination of technologies can significantly increase your ability to transform vast quantities of unstructured data into workable insights. Discover more in our augmented analytics guide.
We've already touched upon machine learning. But how does a machine actually communicate what it has learned? This is where natural-language generation (NLG) comes in: a process for translating a machine’s findings into an understandable format.
Just like Google, ask the system a question via the search bar, and its algorithm will crawl the data to instantly deliver an answer.
Thanks in large part to advancements in deep learning, systems now have the ability not just to read language, but to interpret its meaning and context.
Sentiment analysis classification tools can process natural language to interpret the underlying sentiment of text or speech (e.g. positive, negative or neutral). This technology is especially useful in social media analytics through ‘social listening’ tools.
Chatbots are applications with the ability to engage in interactive conversation using text or natural voice. Used most often in the field of customer service, chatbots use AI-driven natural language processing and speech recognition to answer questions and solve problems. At their best, chatbots are able to replicate the performance of top-performing customer agents.
This is another significant application of deep learning technology. Using a technique known as computer vision, machines have the ability to interpret the content of images, including charts, tables, photographs, video and text. These capabilities are already being put to work in areas such as research & development and healthcare, including the ability to interpret technical drawings and clinical scans.
To help you understand the importance of AI from a business perspective, the phenomenal growth of chatbot technology provides a useful starting point.
By 2020, about 80% of enterprises will use chatbots. Equipped with this technology, businesses are much better equipped to deliver round-the-clock customer engagement and problem-solving. Those without it are unable to offer the same customer experience. In comparison, the value proposition offered by non-adopters starts to look considerably weaker.
Along similar lines, take the example of a retailer who puts augmented analytics and sentiment analysis to work. Compared to its rivals, this company is much better able to spot trends, to anticipate demand and present exciting and relevant product offerings to its customers.
And then there’s the forensic investigation facility that adopts computer vision technology. Armed with this, its clients can expect accurate reports weeks earlier than anything offered by its competitors.
To stay ahead of the competition, to optimise the use of your resources - and to deliver the best possible customer service, the business case for AI becomes impossible to ignore.
Data analytics involves taking potentially huge amounts of raw data and interrogating it to identify patterns. In this way, businesses are able to track key metrics, measure performance and see long-term trends. Basic data analytics lets you see what has happened and what is currently happening across your organisation.
The power of AI takes this a step further. Using “What if” modelling, it uses historical data to predict what might happen in the future. This kind of advanced analytics is not new. Traditionally however, it has always relied on specialist technical input to query the data, to create models and test assumptions.
But once you add machine learning into the equation, advanced analytics is potentially transformed. Through it, the machine itself is able to make assumptions, to test those assumptions – and learn from the results. Best of all, it is able to achieve all of this at a speed that no human analyst could match. For your business, this means being able to test scenarios and predict the consequences of changing variables –
AI in practice
Take the example of a corporate finance department. Under its legacy software, certain simple tasks were automated. For instance, if a client was 30 days late with invoice payment, then a reminder message was automatically generated.
With its new AI-driven systems, the business is much better equipped to identify and respond to potential client credit-risk. This includes the ability to identify the drivers of credit issues and to segment clients based on likelihood of default. The business now uses this information to make informed choices on when to extend credit lines and when to request payments on account.
Optimise pricing by predicting customer reactions to price changes
Lead classification: score a sales lead’s likelihood of closing
Identify the product/service attributes that are most likely to drive sales
Analyse vast quantities of social data to assess market perception of your brand
Segment customers and leads to assign customised marketing campaigns based on customer characteristics
Predict customer churn
Power chatbots with the ability to understand the context and address nuanced customer enquiries
Provide an interactive self-service knowledge base for customers with the ability to direct users to specific content based on real-time analysis of their on-site behaviour
Detect potentially fraudulent credit card transactions
Forecast inventory levels and plan your stock reordering schedule
Predict the impact of market/economic/political variables on your bottom line. Examples include real-time currency exchange rate variations and shipping delays
Provide a data-driven framework for onboarding by predicting the drivers of early-stage staff attrition
Predict interaction volume in customer support centres to inform staffing decisions
Predict long-term recruitment needs based on analysis of staff profiles
On the one hand, there’s the doomsday scenario, with predictions suggesting that as many as 40% of the world’s jobs will be replaced by a combination of AI and automation over the next couple of decades. Against this backdrop, if staff fear that the threat of obsolescence is just around the corner, they are hardly going to welcome the arrival of AI technology with open arms.
The reality is a little more complex - and the idea of a ‘robot takeover’ almost certainly misses the point. For most organisations, the biggest impact of AI will not be on job numbers but on job content.
Accenture recently found that the majority of workers have a positive view of AI. After all, if ‘asking Alexa’ is second nature at home, then it’s natural to embrace the type of technology that makes work easier.
Much of machine intelligence in the workforce is focused on taking care of routine tasks. Other technologies (predictive analytics, for instance) involves interrogating vast quantities of data and delivering insights to the user. Less drudge and more decision making - along with much greater scope to add value to the organisation: this is the promise of AI. It helps to explain why 62% of workers believe that intelligent technologies will create opportunities for their work - rather than rendering it redundant.
For a seamless adoption of AI, organisations need to focus on highlighting these opportunities. Coupled with this is the need to assess existing internal skills and to develop upskilling strategies to ensure that these opportunities can be taken up.
The procurement manager
With the rise of the online marketplace, procurement managers can often face a huge choice of potential suppliers. The challenge is to choose the ones that are likely to deliver the most value to the organisation. With AI-driven search capabilities, it makes it much easier to define and identify the most suitable suppliers.
Ai can help streamline contract management, too - in a way that goes beyond the capabilities of basic accounts software. For instance, through deep learning and image recognition capabilities, systems can learn how to process non-standard invoices and even support the drafting of contracts, including suggesting appropriate conditions and clauses.
The in-store sales assistant
Armed with AI-enabled devices, shop assistants can potentially have a vast amount of hyper-relevant data at their fingertips. This may include real-time information on inventory, orders and returns - and even things like customer reviews, frequently linked purchases and what people are saying about particular products on social media. A smart search facility means that those staff are able to interrogate this data with ease.
When they field queries from customers on the shop floor, staff are now much better able to provide informed answers and offer relevant suggestions.
The engineer in the field
For on-site operatives in areas such as construction and engineering, AI-enabled technologies can help transform job-specific data touchpoints into valuable business intelligence. Through a combination of deep learning and image recognition, observations, findings and imagery can be captured, shared and reported instantly to the back office.
Meanwhile, for senior management, AI can help give you a real-time view of everything that’s happening on the ground across multiple sites. This makes you much better able to monitor potential project delay risks, keep track of sub-contractors - and ensure nothing is missed at closeout.
Successful AI deployment demands clarity on what your organisation is seeking to achieve and the specific business problems you need to solve. Only once you have defined this can you begin to identify the areas where AI could make a difference.
Machines “learn” based on the data they process – so the quality of that data becomes a vital consideration. Instead of stipulating set rules for the software to obey, deployment of machine learning capabilities requires you to provide examples of real data. With AI, the principle, “garbage in, garbage out” becomes even more relevant.
From helping you to choose the right technology through to ensuring data hygiene, integration with existing systems and full training for your people, MHR Analytics is fully equipped to help you on your AI journey. To see what’s possible, speak to MHR Analytics today.