We have to stop using “ethics” as an excuse to avoid investing in AI. Ethical standards are something enterprise leaders must lay out for their enterprises regardless of technology investments.
For example, what DEI standards are acceptable, and what biases does a company want to set in stone? This dictates how AI can ultimately be governed and also which partners in the ecosystem a company should work with, as most enterprise leaders want to be aligned with other like-minded enterprises. For example, if a company deems it important to create a genuine gender balance within its management ranks, it will likely prefer to work with partners who share and practice those values.
Ethical fear-mongering threatens to kill off the commercial gains of GenAI at birth in three-quarters of global enterprises
Three in four CEOs believe those with the best GenAI will obtain a competitive advantage. Yet the data – presented at an IBM event in London covering GenAI and HR – also shows three in four CEOs are willing to forego the commercial benefits of GenAI over ethical concerns.
These ethical concerns are regularly cited as the cause of delays in the implementation of AI and GenAI projects. We hear this often from service providers. Many report an uptick in the volume and value of GenAI projects in Q4 of 2023 – but they also lament how many enterprises are dragging their feet over governance concerns.
Lumping ethics in the same governance bucket as accuracy and transparency has confused enterprise buyers
But many tech firms and service providers have done themselves no favors by lumping ethics in the same ‘AI governance’ bucket as accuracy and transparency. In doing so, they have muddied the waters. Ethics, accuracy, transparency, and openness are fundamentally different.
- Ethics are a reason for governing—the why. Ethics are standards by which an enterprise chooses to be held accountable.
- Transparency is what is required to understand how well those standards are met.
- Accuracy is a measure of AI performance.
Yet we throw these three together and then wonder why the enterprise stands back, confused.
AI can’t be intrinsically ethical or empathetic
Transparency and accuracy are intrinsically machine capabilities. AI can be accurate. AI can be transparent (this is more an ambition than a reality currently). AI can’t be intrinsically ethical like a car, a washing machine, or a gun can’t be ethical. AI can be no more ethical than it can be empathetic (automatically firing out soothing phrases because you have been trained to do so is NOT the same as being empathetic.)
Only humans are capable of devising and living by ethics.
Ethics are not set in stone either. They are highly context–dependent. Context is another reason why leaders should separate ethics from things that can be built into the machine (such as accuracy and transparency). Today’s ethics are not the same as those of 50 years ago, and no doubt not the same as those of 50 years in the future. Hard-coding ethics into AI could prove an extraordinarily arrogant and risky thing for any human to attempt today.
Ethics remains the C-suite human concern it always was – don’t use it as an excuse to delay a tech project
As discussed in a LinkedIn article in 2018: “The challenge when trying to set rules for behavior is the huge cultural weight shaping our view of wrong and right. That view varies from culture to culture and through time.”
Ethics are not for sale. They should not be sold as part of AI governance. The enterprise owns them.
Separating the two reveals ethics is much less an AI concern and much more the C-suite human challenge it always was. Leaders should certainly NOT use ethics as a reason to delay benefiting from the 10-20% boost to business performance our report GenAI will re-shape business economics, identifies.
The Bottom-Line: Separate ethics from legal and regulatory compliance to fast-track your GenAI route to better business performance.
Enterprise leaders should own ethics. They should not leave goal setting and targets for this to third parties – machine or supplier. Leaders should assess the outcomes of using AI against enterprise-owned targets. But AI can only ever be ethical by rote, meaning ethics is one loop humans must continue to own.
Legal and regulatory obligations aren’t ethical concerns; they are compliance issues. Service providers can and should help with these – building in accuracy to measure compliance and transparency to show how compliance is met.
Enterprise leaders should separate ethics from legal and regulatory compliance to fast-track their GenAI route for better business performance.