Artificial Intelligence for All, Responsibly
As we look towards the turn of another year, it is a time of introspection for many organizations. Annual forecasting, budgeting, and planning efforts are likely in progress by now, often bringing increased stress and anxiety to the C-Suite and the managers that report to them. Everyone is asked to predict the future based on entirely insufficient information. No one knows what will happen, so there are usually no alternatives to good old-fashioned guessing. Sure, you can build a spreadsheet, do elaborate calculations, and ultimately convince yourself that your forecasts are spot-on. Still, overwhelming research shows human judgement is prone to error (bias and noise). So, how do you make the best guesses you can? What tools and technologies can help you guess better?
Enter Artificial Intelligence (AI). The availability of virtually unlimited and relatively inexpensive compute and storage capabilities in the cloud today, along with advances in both the capabilities and usability of AI software, offer unprecedented opportunities to ‘crunch the numbers in new and different ways. We are now well past the point where computers are more effective than people at many data-based tasks, including searching, extrapolating, interpolating, predicting, and even winning strategy and video games. In other words, computers have gotten much better at guessing and making judgements than we are!
In the not-too-distant past, designing, building, testing, and delivering working solutions based on artificial intelligence and machine learning (AI/ML) was complicated and expensive. Fortunately, state-of-the-art. AI/ML technology has been evolving rapidly over the last few years, and several major vendors now offer powerful tools and capabilities at price points that are reasonable for nearly all organizations. In fact, according to Deloitte’s State of AI in the Enterprise survey for the 2022 edition, 94% of the 2600+ business leaders surveyed agree that AI is critical to success over the next five years. A recent Accenture survey of over 1,600 C-suite executives and data-science leaders from the world’s largest organizations found that nearly 75% of companies have integrated AI into their business strategies and reworked their cloud plans to achieve AI success. While adoption does remain somewhat higher in the enterprise space, a recent report from McKinsey showed that 56% of all respondents report AI adoption in at least one function.
Microsoft is one of the leaders in data science and management and has invested heavily for years in expanding AI/ML capabilities across many of its products and platforms. The Microsoft Azure cloud offers various services and tools that make it relatively easy to create robust AI/ML solutions for many scenarios. Microsoft has taken the integration of these technologies with core products to a new level with the recent introduction of Syntex, which brings ‘Content AI’ into SharePoint. It enables productivity-boosting functionality, including automatic document recognition, classification, summarization, translation, automated content generation, automated document assembly, AI-powered search, and active dynamic security/compliance/resilience.
In addition to AI-based content processing, there are a lot of other potentially compelling scenarios for many types of organizations, regardless of size, industry, business type, etc. Some of the most common examples of successful AI/ML outcomes include the following:
- General decision support – Predict likely effects of various options
- Interpreting unstructured content – Search, identification, organization, reuse
- Automated document classification – Assign content types, security levels, etc.
- Auditing, compliance, eDiscovery – Understand what you have, where it is, how it is used
- Resume analysis – Interpret and rate candidates based on machine learning
- Anomaly detection – Identify outliers, unexpected results, and process bottlenecks
- Customer churn reduction – Identify at-risk customers in time to take corrective action
- Support process optimization – Analyze support tickets, response/resolution times, issue types
- Language processing – Interpret and translate between languages, natural language queries
- Image recognition and classification – Identify and tag pictures and graphics
- Audio and video transcription – Turn sounds and images into text
- Content generation – Create content and media based on similar existing items, text-to-speech
However, all the “magic” that AI might seem to promise has a critical caveat. Along with the exciting possibilities that AI/ML offers, it is essential not to lose sight of the potential risks that can arise if care is not taken in implementing solutions. Artificial Intelligence can deliver substantial benefits to organizations that successfully leverage its power. But, if executed without attention to ethical, legal, and other people-centric and societal considerations, AI could also cause significant or even calamitous damage to a company’s reputation and prospects.
To ensure maximum value while minimizing the corresponding risks, it is helpful to have a ‘frame of reference to assess AI technologies. In that spirit, Deloitte has developed a valuable paradigm for evaluating AI-based solutions. The six dimensions in the Trustworthy AI™ framework are:
- Fair and impartial – Help participants understand how their data can be used and how AI systems make decisions. Algorithms, attributes, and correlations are open to inspection.
- Transparent and explainable – Assess whether AI systems include internal and external checks to help enable equitable application across all participants.
- Responsible and accountable – Put an organizational structure and policies in place that can help determine who is responsible for the output of AI system decisions.
- Robust and reliable – Confirm that AI systems can learn from humans and other systems and produce consistent and reliable results.
- Respectful of privacy – Respect data privacy and avoid using AI to leverage customer data beyond its intended and stated use. Allow customers to opt in and out of sharing their data.
- Safe and secure – Protect AI systems from potential risks (including cyber threats) that may cause physical and digital harm.
Whether your business is a Fortune 500 firm or a small organization with limited resources, there are almost certainly one or more areas where Artificial Intelligence / Machine Learning solutions could significantly impact the bottom line. Of course, given some recent survey results, you can be sure that other organizations in your space are also exploring the big, beautiful, and slightly scary world of AI/ML. Don’t be too late for the party but be careful on your way there!
Copyright ©2022 by DivIHN Integration Inc. | [email protected].