It is hard to face the increasingly possible reality that, one day, computers will be "smarter" than we are. As a result, many people experience a subconscious cognitive resistance to accepting the value of gaining proficiency in AI tools. However, even if these tools are extraordinarily better at processing bulk data and other tasks, that just leaves us more room to act as thought executives, outsourcing most work to AI and deciding what we do with it. Anyone with access to the internet can learn the skills relevant to this synergistic workflow—and we must begin to measure individual proficiency to positively reward it, and give everyone a chance to demonstrate their future potential.
It used to be critical to know how to do long-division or solve integrals by hand, or write a structured essay for school assignement. Now, you can ask AI models to do this, and present you various options to chose from or refine further. Just as few people find themselves needing to do previously unautomable tasks manually, such as long division, the same will soon hold for even bigger works of mental effort. That doesn't mean we don't have work to do—it is simply more metacognitive than previous technological innovations have called for. In other words, it will be increasingly important to evaluate the worth of ideas over merely originating them.
We can become more creative by leaving other thinking to AI —and retain a niche that works complementarily to how AI models think.
We can begin to measure your individual "collaborative intelligence potential" (CIP) by using AI to approximate how far you get from objectives to outcomes in aggregate conversational transcripts, on a rolling basis that respects the fluidity of human intelligence and recognizes individual growth. Any such analytical techniques can be continuously refined, baked into generative AI platforms as an opt-in feature, and vetted and debated openly.
By assigning CIP scores to individuals, hiring processes may become simpler and more robust. These scores can be multi-dimensional (think: how well a student uses AI to learn, an engineer to process data, a physician to consider differential diagnoses) and accelerate more optimized job placement based more on merit than pedigree.
These tools are here to stay, and this approach is not limited to a single framework—different entities can apply their own strategies for defining what constitutes productive AI-synergistic workflows and individuals can elect which to pick from.
If substantial alignment issues arise, this concept could be very dangerous, as it would enable a simple approach for identifying which people are less “dispensible”. Any implementations would need to consider this with extreme caution. However, there will likely be a longstanding need for physical work that is immeasurable with these metrics, and having this system be elective would ideally further attenuate the possibility of such issues
Schools must immediately pivot to teaching these kinds of skills not as an afterthought, but as the fundament of thinking for future generations.