The Hard Questions
The easy questions can be Googled. They can can be sorted and checked for veracity. Sure, it’s not easy to catalog the answers or find, link to them, or be able to parse natural language queries to find the answer, but it’s doable and it’s been done, for the most part.
The hard part is the hard questions. The questions faced by us individually. These are usually choices between scenarios that have uncertain outcomes. Sometimes, these issues are dilemmas, where any choice has undesirable consequences.
For these hard questions, there are advice columns in the newspaper. There’s Quora and Reddit. There are Facebook groups and even old fashion support groups. Ultimately, however, these are just more datapoints to consider in decision making.
At some point, when we’re indecisive, we might just flip a coin as another datapoint. Or, we’ll consult the 8-ball. However, another advisor is on the horizon. Yup — the ‘ole GPT3 and other generative AI. If Google is used often for difficult questions that sometimes end up as evidence in criminal trials, generative AI services will be asked much harder questions. As different AI services emerge, they might end up becoming an “advising committee” for the individual. Is there a consistent answer among the group? Is there some logic or insight?
While many are using these tools today for entertainment purposes, e.g. “look at this cute poem it wrote”, some will likely be using the tools for asking much deeper questions. This may be assigning unnecessary powers to the technology. This isn’t new — we assign powers to totems all the time, but unintended biases might end up pushing people who are prone to less desirable decisions.
The irony is that this new technology is generating its own hard questions.
