The unprecedented pace of AI development has taken many people by surprise and has added to a growing unease about this new technology. Fears about the speed of AI development hit a crescendo on March 22nd, 2023, when an open letter signed by tech experts and entrepreneurs, including Elon Musk, urged all AI labs to immediately put a six-month pause on all development and testing of AI technologies that are more powerful than the newly launched ChatGPT-4. The letter began, “AI systems with human-competitive intelligence can pose profound risks to society and humanity, as shown by extensive research and acknowledged by top AI labs.” At the time of publication, signatures on this open letter totaled 21,537.
While we believe that the potential benefits of AI technologies greatly outweigh the negatives, we also acknowledge that there are some significant shortcomings of the technology as it stands now. Here is a list of some of the pitfalls of AI that we are currently urging our clients to take into consideration.
Bias – When it comes to AI, what goes in must come out. It’s come to the attention of researchers that the vast amounts of data that AI is consuming can contain racist, misogynistic or ableist information. This article on Dataconomy gives some good tips to help avoid objectionable outcomes.
Ethics – The ethical concerns raised by AI advancement range from privacy, discrimination and plagiarism all the way to existential angst and the question of singularity – the point at which AI will evolve past and overtake human intelligence. A moral compass is imperative as we create and integrate AI systems.
Incorrect Information – Currently, one of the shortfalls of AI is that it often gives erroneous information. Depending on what type of data you’re using or generating, inaccuracies can lead to poor outcomes and angry customers. At this stage, AI still needs to be fact-checked to avoid content that’s incorrect or untrue.
Liability – AI has been known to liberally “borrow” from content it finds online, especially artistic styles and images, without the original creators’ consent. This leaves companies vulnerable to litigation like the current class-action lawsuit in San Francisco against Stability AI, Midjourney and DeviantArt.
Security – There is a concern among large companies like Microsoft and Amazon, that their employees will accidentally share confidential information and proprietary code with an AI tool. AI isn’t capable of signing an NDA or understanding what can and can’t, or should and shouldn’t, be shared with others.
If we can successfully navigate the valid concerns about AI and commit to being mindful with AI research, development and implementation, this evolving technology can absolutely be a net positive force for businesses, consumers and society at large. After all, AI is really only as good or as bad as the algorithms with which it’s programmed, and for the time being, we humans still control those.
If you’re a business leader standing on the precipice of this brave new world, contact us. Mythos Group can help you assess the benefits and costs of AI integration, and we can work to implement a seamless AI transformation in your organization.