As AGI comes knocking and we spend the next few months (years?) contemplating just how potent is this force behind the door, this blog is our way of thinking aloud and sharing what we’re learning from builders and operators on the frontlines. An observation basis the last few GenAI mixers I’ve attended is that there’s more tinkering on how to optimise workflows and – at this point – less blue sky thinking on what the business applications and implications are.
In which areas would new companies built on a new stack be better suited to deliver 10x user experiences vs the incumbent way of doing things? Which use cases are perhaps adequately solved by traditional ML and anything more complex is overkill? What are the ways in which talent reorganises in response to this new way of doing things? Of course, more questions than answers at this point, but as ChatGPT has just reinforced for all of us — asking the right questions is perhaps the most important first step. I’ll be using this space this week to talk about questions I’m thinking about, starting with…
The beginning of the end of the communication premium? For the longest time, people who communicated better have had a clear leg up at work. Whether you’re an engineer or a business owner, knowing the right things to say and the right way to say them was always an accelerated path to your desired outcomes. This has been an obvious source of frustration for the majority of the world’s population that didn’t grow up English-native but were expected to now compete in an increasingly globalized economy. Just getting a foot in the door is extremely difficult without having the right resume, right cold email, right business pitch deck. And Generative AI levels this playing field and how. Incidentally, as I write this, a friend pings to tell me about the guy sitting next to him at a coffee shop using chatGPT to draft a business proposal (“$20 is nothing, I’m happy to pay $40 for it!”).
We see companies building these point solutions as well as target more fundamental long-term skill-building. Not everyone looking to move up in the workforce can afford personalised English training but when the cost of personalised tutoring and language practice crashes thanks to LLMs, the addressable market expands significantly. While a clear consequence of all this is an even greater leveling of the playing field in terms of employment and gig opportunities, the second order impact is likely on baseline skill level across roles. As style gets commoditised, substance is the de facto next battleground. In a job market and consumer culture where comfort with and fluency in English is strongly tied to personal confidence, I’m excited for the mindset shifts this could open up.
We’re still in very very early days of learning and exploration in this brave new world. The trajectory of AI assistants reminds me of the roles inside consulting firms — you start off as an analyst/associate where you basically follow the specific instructions you are given, then progress to a senior associate who is given an objective but has to figure out the corresponding tasks and resources themselves…all the way up to the partner who actually recommends you the problems you should be working on, with or without a prompt. In the last 7 days, chatGPT has helped me understand vector databases, figure out winning strategies in a new board game we picked up, and given (inaccurate) advice on the right time to add milk to my chai. In the next few years, I expect to be told whether or not I should be spending time on vector databases, if there are more interesting board games I should be playing instead of the one I picked up, and be pushed back on my 5th cup of chai for the day. Ready or not, here we go!
Authors
