With great power…: AI development requires ethical use of data

AI = Data use. Data use = Value. This is unquestionable.

Breakthrough innovations in AI together with increased digitisation of services post-pandemic have increased such value, and governments are increasingly attempting to harness its value both politically and financially. 

The UK government’s AI Opportunities Action Plan intends to ensure that AI is “mainlined into the veins of the nation”, with multibillion pound investment planned within the UK. However to be a true and enduring success, AI must be used ethically.

All businesses are now tech businesses whether they realise it or not. AI developments and data use are at the forefront of many industries, AI tools now being used to assist with customer acquisition and product development; pricing and risk management; provision of goods and services; and fraud detection. 

However, as with all ground-breaking developments, the downside cannot be ignored. These range from lack of transparency to unavailability of goods and services due to discrimination, cyber poverty and cyber liability. Without ethical use of data and AI, financial benefits may come at the expense of societal well-being.

Business Interest

Most businesses are data intensive.

This has inevitably led to the increased use of AI in order to benefit businesses and customers. We have first-hand experience of the importance of data to organization, having advised on the creation and management of major strategic databases including the MID, CUE, MyLicence, IFR, MIAFTR and ELD. Many have involved complex relationships with third parties (e.g. DVLA) and are underpinned by ethical processing.

Businesses also need to be trusted custodians of personal data; that is crucial for customer confidence. Therefore, in addition to regulatory obligations (which the UK government may now soften), such valuable data assets must be protected and used ethically.

Get it wrong in relation to AI and personal data, and a potential PR disaster looms: reputational damage and loss of customer confidence are often more damaging than a regulatory fine or sanction. 

All businesses should view ethical use of AI and personal data as paramount. Future AI innovation and other advances in robotics and automation promise new opportunities to improve business efficiency and, importantly, the customer journey, their interests and outcomes. Ethical use of AI and customer data needs to be the next step in customer service.

Customer interest, coupled with the underpinning of ethics within an organization, will benefit society as well as institutional stakeholders. Therefore, in addition to assisting in relation to the organization’s CSR and ESG objectives, for example, ethical use of AI and data makes business sense, too.

Public Interest

Public interest is another important consideration in relation to any organization’s use of AI and data.  When deploying AI systems, business must be aware of the risks that model outputs may mirror societal biases in the training data. This in turn risks discriminatory and unlawful consequences, particularly where systems are used to make or support decision-making.

Furthermore, the 'black box' nature of AI systems presents significant challenges for organizations subject to legal obligations of fairness and accountability where AI is used in decision-making. Careful consideration will need to be given to the lawfulness of reliance on opaque technologies in decision-making. 

The above are often competing considerations during the operational lifecycle and thereafter. However, ethics and accountability should underpin all decisions in relation to AI and data usage.

Training Data IP and Copyright

The models underpinning generative AI are trained on vast data sets. Often, the training data includes large amounts of material scraped from the internet, which includes the creative output of authors and artists, and a range of other material protected by intellectual property rights.

The commercial exploitation of that data without the creators’ agreement is a growing point of controversy, which will take time to play out. There are a number of high-profile lawsuits in progress in which creators are seeking recompense for what they see as the theft of their work. It will take time to see how the laws in various jurisdictions resolve these questions. 

How Do You Achieve Ethical Use of AI and Data?

Transparency

Arguably, this is the fundamental principle of ethical use of data. Be open and honest about your use of AI and data; whether via Privacy Policies, consents and or approvals.

Understand your own decision-making in relation to data use – if you do not fully understand it, how can you explain it to customers, investors, stakeholders and (potentially) the ICO?

Understand your customer

If you do not view AI and data use through the lens of your customer, it is difficult to gain and maintain trust and confidence. Interrogate your systems and processes to ensure that your customers’ needs, concerns and expectations of data use are anticipated and applied.

Respect customer choices

If you rely on customer choice (consent) to use their personal data, you must make sure that choice is informed, freely given and can be withdrawn as easily as it was given.

Subject to law and public interest, you must restrict the use of a customer’s data to that which is fair and/or expected, particularly if automated processes are applied.

Maintenance and integrity

Equality and fairness are the foundation of ethics in this arena. Identify and remedy data sets that could bias decision-making.   

Cultural identity 

Ethical use of data should be engrained from the Boardroom to the ‘shop floor’. Ethics should be built into decision-making at all times.  

Be accountable

All should be accountable for the ethical use of data, with a defined chain of command. Accountability should also permeate the supply chain.

Earn and maintain trust 

Firmwide and consistent application of the above principles should go some way to meet the organization’s commitment to ethical use of data. A valuable by-product will be the acquisition and maintenance of customer trust – a commercial advantage, which cannot be bought.

Long-term Value

Compliance with the law is a given for all businesses, however ethics are as important in relation to the use of AI. Data is a valuable commodity and, if used correctly and ethically, can assist businesses considerably. However, organizations are also caretakers and guardians of the personal data they hold - without ethical use, it is unlikely to be in a position  to utilise that valuable asset in the long term.

[Photo by Claudio Schwarz on Unsplash]