What skills do AI professionals need to address social and environmental issues?

Worth mentioning that AI is not just onething but it is rather based on several disciplines. John McCarthy who was the first to coin the term AI in 1956 defined as “the science and engineering of making intelligent machines.” Most of the algorithms used in AI are deeply rooted in mathematics and statistics. So one way to improve your technical skills is to enhance your knowlege in the STEM area.

For AI professionals developing models: - be aware of the environmental impact of the infrastructure used in the design - be able to architect solutions that are 'Green by Design', minimising the environmental footprint, and with automation to optimise usage and scaling of storage and compute, if necessary with hard constraints aligned to the organisations ESG goals. - make recommendations to decision makers of the ESG impact of one model / approach vs another

AI ethics and bias training are some of the most in-demand skills right now. Most companies and individuals can deploy AI systems with little to no oversight but it will soon not be the case. AI professionals need to become aware of the level of bias present in most datasets and the societal implications of it. Most datasets hold bias due to historical reasons, and it is key not to perpetrate discrimination through algorithms trained on faulty data. It is particularly key to go beyond high-level principles of fairness and equity and actually learn about pragmatic ways to mitigate and counteract bias.

Some days it feels like AI sells itself. But as the number of models and AI-driven services expand rapidly, being able to communicate effectively about _why_ AI should be used (and why a particular solution would be best) becomes critically important for AI professionals. Otherwise, others will fill that void and might steer decisions in a different direction.

Running large models at scale generally requires significant compute power, which in turn usually requires significant energy demands. It could be considered irresponsible for AI teams to focus solely on the application of responsible AI principles to the discrete use cases of AI-powered services without taking into account the potentially significant externalities that can be associated with at-scale deployment. This is an area where cross-enterprise coordination (between, for instance, ESG leadership and engineering leadership) can facilitate best practices and risk reduction.

Critical skills are key in particular questioning the value add of AI. My recommendation is to never start with AI but always start by framing the problem/opportunity first then bring AI as a tool to solve it.

Creative Skills are amalgamation of Innovating Ideas, Out of the box thinking, and Communication Skills. Others Skills such as Empathy and problem solving etc can also be part of creative skills.

To address environmental issues, AI professionals should possess a solid foundation in data science, machine learning, and modeling techniques. They also need a strong grasp of environmental science and domain knowledge to effectively apply AI in areas like climate modeling, resource optimization, and pollution monitoring. Collaborative skills for partnering with experts from various fields and the ability to develop innovative, eco-friendly AI solutions are key to making a meaningful impact on environmental challenges.

In practice it is rare to find individuals that have all the above skills. Whilst the emphasis here is on individuals approaching the topic from a team perspective makes sense. If there are gaps at individual levels it is fine as long as the collective team skills cover all the above.

Cultural sensitivity is critically important. AI solutions are deployed globally. Being culturally sensitive ensures these solutions respect local traditions, values, and norms, and avoids unintentional negative consequences. For example, - An AI-powered translator recognizing and accurately translating non-gendered pronouns from languages that have them, such as the Swedish "hen". - Ensuring that datasets used to train facial recognition models are diverse, representing a broad spectrum of ethnicities to prevent biases. - AI tools used for content moderation on social platforms being trained to differentiate between culturally specific content that might be deemed offensive in one culture but acceptable in another.