The economic, social, and environmental benefits of AI can only be fully captured by applying a rights-based approach to ensure responsibility.
We’re motivated to make AI work. That’s exactly why we need to embrace the human elements that judiciously balance right versus wrong. CIOs have a moral obligation to weigh the needs of the organization with a rights-based approach to ensure the ethical use and adoption of AI.
AI has moved from simple search algorithms to integrated statistical analysis. As expert systems move beyond symbolic logic instead of defaulting to only generating numerical calculations, these systems can extract knowledge from explicit content repositories. This ability to interpret conclusions creates decisions that are understandable to users.
The near-magical fascination with expert systems quickly refocused energy on the competitive advantages of decision-support systems that are able to apply concepts across healthcare, financial services, genetic engineering, and air-traffic control. However, the challenges of expert systems have been many. Software standards and interoperability, knowledge acquisition and analysis, handling uncertain situations, system integration, validation, and plain ol’ user-adoption resistance has slowed the realization of AI. All of these challenges are real.
No challenge presented by AI has been more overreaching than designing a rights-based approach to ensure responsibility.
A rights-based approach for financial services
Machine learning (ML) and deep learning (DL) are rapidly transforming the global financial-services industry. The new physics of financial services has been reborn through opportunities enabled by AI. The six, main financial sectors benefiting from AI are:
- Investment management
- Capital markets
- Deposits and lending
- Market infrastructure
Investment management is using AI to mimic advanced strategies while controlling costs, finding new and unique correlations among datasets, and analyzing vast quantities of data at scale. Capital markets have achieved better investment performance by using new data in opaque markets, developing real-time insights, and for pre-and-post trade-risk management solutions. We’re seeing payments automate compliance and reporting, act as the ultimate personal shopper, and create an advisory capability for macroeconomic trends. Insurance can process claims instantly, develop modularized policies, and advise clients on prevention strategies to lower their risk exposure based on past actions. Deposits and lending have the untapped potential of applying predictive algorithms and estimating defaults with greater accuracy, miniaturizing unsecured lending to be use-specific, and providing true, just-in-time lending. Market infrastructure is exploring selling internal analytics capabilities “as a service,” automating alert triage and investigative reporting, and integrating post-trade workflows to achieve straight-through processing. Combining finance and AI has the potential to change the physics of financial services in the near term.
From “doing the same thing better” (A) to “doing something radically different” (E), AI opportunities can be classified into five strategies:
- Leaner, faster operations
- Tailored products and advice
- Ubiquitous presence
- Smarter decision-making
- New value propositions
Applying these strategies as you identify and evaluate AI opportunities can assist your leadership teams in quantifying the organizational disruption of adopting a new operating model wrapped in AI.
AI is relevant for financial services and can be applied in a multitude of ways:
- Operations and management
- Services delivery
Not much is more personal that your money or the method by which you gain access to it. The right for humans to do this varies by industry segment—whether we’re discussing assess management or investment banking.
Are you approved for a new credit card? AI might be behind the decision. Did your insurance rates increase? AI might be to blame. Poorly designed algorithms could result in programs making decisions on discriminatory factors such as nationality, social status, usage, or spending habits.
Also, many Millennials are steering away from picking individual stocks and, instead, choosing funds that automatically rebalance. Electronically traded funds (ETFs) are investment funds that typically operate with an arbitrage mechanism designed to keep them trading close to their net asset value. AI has big opportunities here, but a concern revolves around the rules that define trade. Is the fund’s goal profit maximization? Is the goal environmental advancement? Responsible investment is a relatively new dimension when applying AI to learn how to trade and invest. If left unchecked, the funds could quickly own pesticides and autonomous weapon systems that have high margins and lower risk but might miss the social goals of the investor.
A rights-based approach for healthcare
The potential for extending life can be done better with AI. AI enables population health. Front-line health workers can benefit from virtual health assistants. Physician clinical-decision support is more durable with AI.
AI covers three major areas in healthcare: data, processing, and action.
- Computer vision
- Speed recognition
- Natural-language processing
Computer vision can use automated methods to conduct image-based inspection and analysis of CT scans and MRI images. Speech recognition identifies responses to sounds produced by the human voice for hands-free patient interaction. Natural-language processing can aggregate large amounts of written data into natural language to develop narratives for patient records.
- Information processing (by AI)
- Machine learning
- Planning and exploring agents
Information processing digitizes data using methods that parallel human thinking to anticipate treatment and care. Machine learning is able to recognize patterns and learn to improve an experience without additional programming. Intelligence nanites or robots can perform surgeries.
- Image generation
- Speech generation
- Handling and control
- Navigating and movement
Image generation can take multiple images from MRI to create a 3D image that’s projectable, presenting a mixed-reality world and better analysis. Speech generation can recreate human-like speech from patients that are unable to generate voice commands on their own. Handling and control use AI to perform the automatic handling of objects, such as disposing of one-time-use products after a surgery. Navigating and movement can transport patients seamlessly to and from their primary residence or care facility to appointments such as checkups or annual exams.
AI will change healthcare. Our goal is to ensure that this change is for the better. AI can be applied to healthcare in these areas:
- Treatment plans and personalized medicine
- Optimizing the clinical process with personal, connected health
- Patient monitoring and assistance
- Research and development
- Connected healthcare systems
Human-rights issues around AI begin even before the patient enters a provider’s office. AI can provide progressive, evidence based-treatment plans and learn from treatment results. Right-to-health disparities can surface when patients don’t have the same access as AI agents do to information, analysis, or patient historical information. Privacy and confidentiality of personal and highly sensitive treatment results demand robust data policies, controls, and data stewardship. As AI creeps into the labs to assist with patient clinical trials, informed consent of what AI can and can’t do is less a matter of AI function and more a matter of what functions the patient has given consent to have AI access.
AI can provide high-quality and efficient care with precision medicine. AI leverages patterns generated from linking huge data sets of genetic information. However, prior to AI providing care or treatment, a deep understanding is required of the sources and credibility of the data sets used in treatment analytics.
Is AI a new dimension in health inequality, or can we, as leaders, design AI for responsible healthcare?
A rights-based approach to AI research
Crowdsourcing citizen diagnostics and treatments (23andme, CureTogether). Deep learning with unlabeled data (Ginger.io, physIQ). Use of legacy health records for predictive analytics (ikioo, BenchSci). Next-generation, personal financial management (claritymoney, MoneyLion). Automation of savings and bill payments (citi, WeChat). Multiple-provider platforms becoming financial managers (credit Karma, tink). Each possible outcome is real. We need to shape these outcomes, so they’re realized responsibly.
AI quickly commoditizes old methods of customization, capturing attention, and developing ecosystems. Customer-behavior insights, product-development experimentation, and new differentiators will be stretched, squeezed, and inspected to get that last drop of value. In that quest for value, keep it human.
Shared AI prosperity requires a deeper, cross-ecosystem collaboration. AI is more than what we do. AI is about changing how we work. How is AI changing your environment?