Artificial Intelligence: demystified at your doorstep

In recent years information technologies made significant progress in the fields of digitalization, data manipulation and analysis which enabled innovative solutions like Artificial Intelligence (AI) to emerge. The development of “smarter tools” for machine learning (algorithms) made mass usage of artificial “reasoning” feasible. AI quickly spread into business and created new markets of potential users for these solutions. In the same time data processing power and transfer speed increased significantly to make cloud computing (data storage and “software as a service”) widely available even for small firms.

The arrival of new tools and opportunities have triggered competitive activities between innovators as well as between early adopters and those sticking to traditional solutions. Consulting house McKinsey noticed uneven progress among companies in adopting these new business solutions. While majority of firms struggle with implementation, not making a quick benefit, those who fully succeeded implementing AI into their business models and processes have reached multiple benefits while transforming their businesses and overtaking and disrupting competitors.

AI is a scientific field of information technology which tries to mimic human decision making. It is often swopped for Machine Learning – a subdivision of AI. Machine Learning uses mass data analysis to learn patterns in cause-effect relationships within “mass chaotic data” in order to come to educated conclusions and recommendations for the future. It can be applied for speech recognition, languages and translation, image recognition, learning, reasoning, concluding, strategic thinking, planning, intuition and decision making. In business AI looks for trends and patterns in mass data to predict failures, successes and other events. It is already here – waiting to be implemented and applied.

In order to start benefiting quickly from these tools one will need a well organised large database. If you are a large company you probably already started gathering data about customers, suppliers and/or any malfunctioning in your processes. Over time you managed to build a pretty large database which AI can now analyse and provide recommendations about how to generate new sales or make savings somewhere in the process. For small firms there are private databases for hire like Google, Facebook or Amazon. Although we witnessed letting or selling parts of those extra-large customer databases to third parties’ AI analysis before, current laws related to privacy and data protection may tend to prevent that from happening in future. However one can still use those three large databases indirectly by paying those companies for the full AI service (use of the database and AI analysis based on supplied parameters) and receive recommendations only. It is well known that sophisticated target marketing which provides successful sales leads makes a lot of money for the trio.

AI enables applications to constantly change/upgrade their own software and thus performance without human interference/reprogramming. These AI programmes possess learning and own code modifying capability based on enough new data which makes them capable to, for example predict future sales based on historical results and current conditions, identify suspect behaviour on Internet and protect from attack, learn to drive, improve process efficiency by identifying waste, rank job applicants, predict a content popularity or recommend best content for a target group.

Marketing and sales are among the first adopters of AI in business. As I said everything starts from a large database about existing purchasing customers, contacts who haven’t placed first orders yet and prospects whose info is in your possession but haven’t been contacted yet. AI helps in identifying those with greater probability to purchase from you (for the first time or again) and recognises patterns in their buying behaviour and decision making routines giving you the opportunity to prioritise those prospects in your sales and marketing activities and approach them in a way their minds are most approachable. Instead of chasing 100 potential buying customers AI may provide names of just 10 most likely immediate buyers on the basis that they are the best fit with profiles of your past clients at the time of purchase giving you the opportunity to make equal amount of sales with just tenth of effort. You will call those 10 guys at the time when they are just a step away from purchasing your kind of goods anyway. Your approach will be personalised with specific arguments which aim to bring his/her positive decision about business with you much closer. In this way a company saves on salesforce workhours, speeds up a sales cycle and improves ROI of these activities.

AI can also help with customer segmentation. It will study behaviour of a large enough population from your database and analyse probabilities and timing of their future buying decisions providing you with an estimate which customer groups best fit your company. How does AI do this? Of course, half a job is in well performed data analysis with a goal of describing your most usual buyer persona and his/her behaviour in such situations. AI creates algorithms that recognise behavioural patterns of your buyer persona which then become the basis for evaluating all database members. The tool looks for similar patterns and predicts future behaviour by estimating who will fit better with your sales offer. Those with more similarity to the pattern score higher and move to the top of your “to do” list. The information also includes predicted buying behaviour for those selected. You can now understand how important and valuable could be data from social networks and your own Internet group.

A company can’t just install and run AI application like any other software package because it doesn’t come ready for its final usage. It still needs to iterate with your database in order to develop its own intelligence i.e. learn algorithms applicable to your data and your specific business case. AI learns and improves itself permanently by iterative testing of own assumptions with positive or negative outcomes. In this way each next round of testing/iteration always starts with improved assumptions that result in better quality of concluding algorithm. In practice AI team would first set root algorithms and then trigger self-improving iterative process with the database. Their ultimate goal is to bring quality of the final product to a viable business application level. This normally happens in-house as AI engineers obtain data from your IT system and use your business logic and not some hypothetical/general values.

Should you build AI application in-house or simply buy it and let installation and implementation to others? On one hand it is difficult to attract experienced professionals to work for you as they are already engaged with AI vendors and, on the other, no one can develop AI without a good quality database and well trained algorithms which both depend on you. The solution is somewhere in between. It is good to start from a database setup: Which data to collect? How to organise them? How to feed algorithms with useful data? After sorting these issues you need to define root algorithms and how to train them in order to get to a point when AI analysis yields usable results. It all depends at what stage you are with digitalization now and how strong is your IT organisation: human, hardware and software resources. AI development is not an isolated endeavour for IT team but rather a strategic decision at the executive level of a company as it usually brings change to your business model, competitors and markets what, no doubt, demands full commitment from the board as well as a lot of extra work.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: