Diverse and inclusive products - legalities, risks and artificial intelligence
“Our ability to reach unity in diversity will be the beauty and the test of our civilisation” - Mahatma Gandhi
As discussed at our recent panel discussion, ‘All for one and one for all – today’s diverse and inclusive products: legalities, risks and AI', diversity, equality and inclusion (DEI) must be taken into account at each stage of the product lifecycle including product design, product testing, and product use.
Designing a product with DEI in mind is about designing for as a diverse range of people as possible. There are not only commercial reasons for doing this but also important ethical and legal considerations, such as:
- Comfort and protection
- Fair representation
Many products across different sectors have, historically, been designed with men in mind, resulting in women potentially being placed at risk of injury and/or discomfort when using such products. Notable reported examples include automobiles using crash-test dummies designed based on the ‘average’ male and hand held technologies, such as a mobile phone, designed based on the sizes and characteristics of men.
As highlighted during our panel discussion by Heather Bramble, Vice President of Product Safety and Regulatory Compliance at Mattel Inc, Mattel has taken a leading role in centring representation with a fashionista line-up of diverse and inclusive Barbies. The line, available from June 2022, includes a doll with a prosthetic leg, a Ken doll with vitiligo, and a range of dolls with a variety of body and skin types.
The testing phase of product development ensures that the product operates as intended. From a DEI perspective, this means testing and evaluating a product with as a diverse range of users as possible in order to minimise bias and any consequential racial, sexual, gender or disability based discrimination.
In the context of AI, various tools, for example, have reportedly misidentified individuals with particular skin tones. Similar issues have also been reported in diagnostic software, sensors on autonomous vehicles and facial recognition technology for airport passenger screening. Potential causes of this include the inadvertent introduction of bias by developers; the use of flawed (i.e. biased) data when ‘training’ AI technologies; and/or testing the product on an insufficiently diverse population.
Research into discrimination by AI technologies is ongoing and organisations are continually looking at ways to mitigate these risks. For example, the UK’s Information Commissioner’s Office (ICO) is developing an AI Auditing Framework to mitigate discrimination risks in AI models. The EU has also introduced a proposal for a regulation on AI that aims to tackle the risk of erroneous or biased AI decision making.
Considering the use phase of the product lifecycle from a DEI perspective is about ensuring that a wide group of persons can effectively use the relevant product and that the manufacturer and their suppliers are aware of the risks where this is not the case.
The continued surveillance and evaluation of products post-release to market and the monitoring of user feedback is fundamental in this respect. It will not only help ensure users’ needs are met but also help manufacturers identify exclusions or inadequacies that may require rectification as well as avoiding potential reputational damage
Application of product safety legislation
Whilst there are general overarching product safety laws such as the Consumer Protection Act 1987 and the General Safety Product Regulations and technical product standards that will be of relevance when considering DEI issues in the product sphere there is currently a dearth of specific legislation in this area.
Particularly in Europe, we have overarching product safety regulations which are so broad in nature that they require technical standards for their interpretation and these technical standards will list a whole raft of testing requirements, but there’s an almost audible silence in respect of diversity, equality and inclusion.
Sarah-Jane Dobson, Partner, London
Pending the expected reform of the UK’s current product safety and protection laws, which is to include consideration of a diverse and inclusive product safety framework, product manufacturers should consider existing legislation and guidance when developing product lines. For example, guidance relating to AI technology published by the ICO, and anti-discriminatory legislation of a more general application, such as the Human Rights Act 1998 and the Equality Act 2010.
As the regulatory landscape continues to develop and evolve, stakeholders working in this area can take certain steps to improve DEI within their organisations. Best practice steps may include:
- Employing a diverse workforce comprising individuals from a variety of backgrounds who bring differing life experiences, views and perspectives to the table.
- Ensuring product designers and staff safe are educated on cognitive bias and how to combat it.
- Undertaking testing with a diverse group of consumers.
- Relying on data sets relating to diverse populations in order to mitigate the risk of biases creeping into product development and ensure they are identified and rectified when they do.
- Ensuring appropriate warnings and limitations are incorporated into transparent instructions for use to mitigate product safety risks.
- Listening to customer feedback and continually assessing user needs.
Ultimately, it is important that both individuals and corporations within the product sphere incorporate DEI principles and implement best practice to minimise any potential risks and liabilities and to improve the consumer experience.
All for one and one for all - today's diverse and inclusive products: legalities, risks and AI
This interactive panel discussion consider today’s diverse and inclusive products, contextualised by relevant legal frameworks and product safety and liability exposure where relevant.Watch now