Share to Facebook Share to Twitter Share to Linkedin Every new advancement has varying degrees of impact on society because it changes the way we think, feel, or operate. This is especially true of AI. Now, AI is not new on the scene, but the last 18 months has seen artificial intelligence supercharged which has meant that it features in more and more aspects of everyday life. Let’s be honest, there isn’t a single industry that hasn’t at least dabbled with AI. What this means is that we now sit with bigger ethical considerations, because no, you cannot launch facial recognition technology for mass surveillance without consent and yet AI can help humans do just that. There is an equilibrium between the efficiency and monumental strides (think medical diagnostics) we all want and human rights being upheld in the most ethical way. What matters now is how we get to that point. H2: Ethical Challenges in AI Development There are so many examples of the negative effect of bias and discrimination in AI algorithms which is why there is so much rising concern around them. Jen Clinehens gave a great example in a podcast last year about Amazon’s HR AI tool that was built by mostly white male engineers. The result was a system that started to favour people from specific schools and of specific cultural background because the people who built it weren’t diverse, the tool itself wasn’t either. MORE FOR YOU House Of The Dragon Season 2 Episode 1 Recap And Review Blood And Cheese Heat Wave Threatens More Than 100 Million With Dangerously High Temperatures This Week—Here’s Where How The Market For Klay Thompson Could Be All Over The Place Unfortunately, the pattern of bias matches societal challenges as a whole which means that marginalized communities often get the worst deal. But the pendulum does swing the other way as well. Nestlé uses AI to build on their understanding of consumer behaviour across different demographics to tailor their campaigns to better suit marginalized groups. Unilever also uses AI to analyse consumer data from various sources to reach different demographics to create products that resonate with diverse consumer bases. AI is as much use as a chocolate teapot without data, the more data it has access to, the better it is as a co-pilot. Of course, the more data flying around to fuel the AI, the higher the risk of breaches. And let’s not forget the aspect of consent, utilizing data without permission is in itself is a breach of human rights. Remember when Clearview AI experienced a massive privacy breach? They created a facial recognition database using over 3 billion photos scraped from social media and other platforms without permission. This database was then sold to law enforcement agencies and some private entities which led to monumental concerns over privacy and multiple lawsuits (obviously). Another case that comes to mind is Google acquiring DeepMind and its partnerships with healthcare systems. Allegedly, Google gained access to hundreds and thousands of patient records (minus the all-important consent) and then used this data to train diagnostic algorithms. The desired outcome is amazing, but the method raised concerns with HIPAA amongst other privacy laws. Basically, AI can help us do some pretty spectacular things but it has to be developed on a human-first attitude which means: · Consent is required for data to be used to train AI · There needs to be transparency and accountability across AI decision-making · We have to develop governance frameworks that are ethical H2: Regulatory and Policy Frameworks The need for regulations has been a hot topic for a while and there are regulations being put in place globally at the moment. H3: USA · Biden’s Executive Order on AI – focuses on making sure AI systems are safe, secure, and trustworthy. · California Privacy Protection Order (CPPA) – has issued draft regulations on automated decision-making technologies under the California Consumer Privacy Act (CCPA) · Federal Trade Commission (FTC) – addresses issues like bias in algorithms and protecting consumers from being deceived or harmed by AI H3: EU · AI Act – regulates high risk AI systems to ensure transparency, accountability, and the minimization of bias · AI Liability Directive – addresses legal liabilities and consumer rights by allowing individuals harmed by AI to seek compensation H3: Global Protection The United Nations is creating guidelines for global AI governance which will require international cooperation to set standards worldwide. UNESCO has established a Global AI Ethics and Governance Observatory to provide a resource for policymakers, regulators, academics, the private sector, and civil society to find solutions for the most pressing AI-related challenges. H2: Technological Solutions for Ethical AI Apart from permission, which has been discussed, bias is another large aspect of correcting algorithms for ethical reasons. Starting with data collection, it is essential that this data is diverse and representative of all groups particularly underrepresented groups. Synthetic data can be utilized to balance datasets if this isn’t the case. Bias metrics such as disparate impact, equal opportunity difference, and demographic parity measure bias. These should be part of every AI to adhere to ethical baselines. When you come to the algorithm, there are techniques under pre-processing, in-processing, and post-processing designed to ensure fairness and reduce bias. Other key points would be to run fairness-aware model training, keep the human in the loop for effective feedback, and ensure model decisions are transparent. For humans to have a comprehensive understanding of AI is crucial. This maintains the power balance and keeps the human at the controls. Using tools that we don’t understand or don’t have visibility of how they are trained or built can be really detrimental specifically when it comes to the protection of human rights. H2: The Role of Businesses and Organizations Tidio found that nearly 69% of college graduates believe AI could take their job or make it irrelevant in a few years. When we’re dealing with this level of fear for livelihoods there is a corporate responsibility within ethical AI because AI is not meant to push people into poverty, or make their lives difficult. Every company has a responsibility to use AI in an ethical way. This is why more and more companies are appointing leaders to oversee AI technology. Some CPGs that have made this step include; Unilever, PepsiCo, Johnson & Johnson, Coca Cola, L’Oréal, and Kimberly-Clark. Employers are also responsible for ensuring their teams are supported with training and awareness, as well as encouraging ethical AI practice across organizations. It’s a full circle commitment. If you want the benefits of AI within your company then you are also responsible for making sure that the AI you implement is ethically sound. H2: Conclusion I have an AI company, so I understand how important it is to show explainability, transparency, and work against bias. AI is moving so quickly that we all have a duty to demand regulatory frameworks that will protect the human through these remarkable transformations. AI is changing the world for the better in most cases, but for this to continue and for people to be protected it is fundamental that ethical considerations are continuously at the forefront of our minds. Because at the end of the day, there is no point in having the benefits of AI if human rights are being assaulted in the process. Follow me on Twitter or LinkedIn. Steve King Following Editorial Standards Print Reprints & Permissions
Subscribe
Login
0 Comments
