A lot has been written about AI and its impact on humanity. Instead of covering what you've read before, we thought we'd focus on the new and less discussed. We'll keep this short:
- The resource: Lex Fridman recently interviewed Sam Altman, the founder of Open AI (the company behind Chat-GPT). In the 2 hour podcast, they cover the company, their approach, the benefits and dangers, what we need to build, what we need to learn, and much more. It's an interesting interview straight from the horse's mouth, recommended: link
- Point 1: A return to nuance. Consider that two of the most consumer-impacting products of the last decades - internet search and social media - have an incentive to show users what they search for. Search engines do a much better job than social networks at balancing their results, but the risk of falling into an echo chamber is significant. An interactive AI has the opportunity to provide context, counter-opinions, and even debate with the user during the interaction. We're very aware that this puts more pressure on removing bias from these systems, a hard problem. Still, a return to more nuanced opinions, more self-doubt, more civilized discourse, is something we'd welcome.
- Point 2: A new way of interacting. Because Chat-GPT provides such impressive results, something that is overlooked is the new user interaction it supports. Consider examples of innovations that dramatically enhanced and/or simplified the interaction abilities of their users: the iPhone, the ATM, mobile banking platforms, ride-sharing apps. We think Chat-GPT also falls into this category: imagine that you want to cook spaghetti pesto and are looking for a special recipe ("give me a recipe for spaghetti pesto with extra garlic"). Before, you would have had to browse recipes until you'd find the right one. Now you can converse with the AI until its generated a result tailored to you ("can I add cauliflower to the recipe?") as well as amend it on the fly ("I accidentally spilled milk into the bowl, how do I save it"). Beyond the accuracy of the results, the very way we can interact with these algorithms is different and a marked improvement.
What do you think is not talked about enough ?