Our AI Event Horizon

event horizon (noun)

1. A theoretical boundary around a black hole beyond which no light or other radiation can escape.
2. A point of no return.

Over the past few months, I’ve shared a few thoughts on the proliferation of (generative) AI. My thoughts on this were(are) always evolving, but in each instance I eventually meandered my way back to this repeated thought: None of these "new" companies (Midjourney, Anthropic, or even OpenAI) seem to have the infrastructure to beat out the big players. With the likes of Microsoft and Google building AI into their widely used networks of software and hardware, how were these smaller start-ups to survive? And now … Apple Intelligence (we see what you did there, Apple).

As if this final Behemoth jumping so loudly into the AI ring wasn’t proof enough that the smaller players couldn’t compete, we also found out that Apple plans to integrate ChatGPT directly into its OS starting as early as this fall. This reminds me of the story of Intel and Apple. Intel had a good 10+ year run of supplying processors/chips for Apple products, but ultimately were cut out so that Apple did not have to rely on outside sources for its products. Just as they started producing and using their own processors in 2020, I expect that after a much shorter run of offering "the current best AI tools" on their devices, Apple will develop(rather, finish developing) their own generative AI and abandon OpenAI. 

… But what do I know?

And this brings me to why I shamelessly clickbaited with the event horizon definition to open here: I believe we’ve just reached a familiar point of no return for generative AI. A point at which the established players will begin to grow stronger in the field with increasing speed (and likely greed and/or recklessness). A point where competition will dwindle, and when something new and promising does arise, it will be bought or copied out of existence by one of these larger players.

I think my over-simplified takeaways here needs to be split between professional and personal considerations.

Professionally, we just need to continue to educate ourselves as things evolve, and proceed with caution. Many of you work in an industry where you are dealing primarily with the IP of your clients (as do I). So, barring a completely local AI tool (a limitation that currently excludes the use of the largest, most powerful tools), we must carefully consider each use of generative AI in our workflows. Especially as it begins to be built into every piece of hardware and software we use.

Personally, we all have to work to understand the application and implications of having the most powerful AI tools build into our personal devices. What are the default settings/what do you have to opt into or out of? What data do these tools have access to? What is/could be shared with third parties, or used in marketing? In short... what could go wrong?

I'll admit that my initial reaction to hearing that ChatGPT was going to be integrated into Apple devices was "finally! My Siri/Alexa/Google home assistant feels so dumb compared to the potential power of a good AI assistant!" But now that it's so close, I feel more cautious. Our phones are very personal devices. They hold photos, passwords, personal documents, health data, preferences... everything (thats potentially not hyperbolic, depending on how much you rely on your device).

So, celebrate! ...But with caution. What a time to be alive! ...But don't take it for granted. Continue being the thoughtful, considerate person you already are... And temper your excitement for new possibilities with an insistance that the (fewer and fewer) companies developing these possibilities, act as thoughtfully and considerately as we do.