I don't think this tweet has aged well for me.
There were some things that I considered when adding "AI" as part of the "dummy/deceitful trend", but I'm pretty sure part of the skepticism was a little bit of fear.
Did I just saw a trend in raise with no clear future? Was I just angry at my previous job and therefore just spitting negativity?
It is now December of the same year, just barely 5 months later and my thoughts on AI have completely changed. I've already made the leap of faith.
From trend to reality
First thing first, lets address my Google bias. I've been following AI progress from Google for a long time, including adding AI into the Pixel phones, their wins with Alpha-Go against humans or their applications of AI in science.
Actually, one of the first things I remember regarding AI, were the dreams that were initially shared by Google Research. I remember thinking immediately "do robots dream of electric sheep?". Now it is clear that LLMs are just dream machines.
But then, my reactions to others doing the same was being skeptical. How did they get into it faster than the ones that have been doing it for years? Well, it turns out, I was trapped in my own echo chamber.
The AI Wars was something that has been cooking for a long time now. And in some ways I wasn't late to it, but in others I was behind the trend.
In the last 5 months, I took a race to update myself and surf the trend again. A lot in the AI field has already changed, including the things that I'm reading about it. Here I share my highlights:
-
Google finally started stepping up in the AI battle with OpenAI, first by launching and improving Bard, and then by unveiling Gemini. I know they have lost the initial battle on marketing AI to the masses, but I've been following Google Brain and Deepmind for a while now, and I really love the way they made progress in the past. I won't stop seeing them as pioneers in the field applied to actual use cases.
-
The Browser Company launched some really good features for Arc (my default browser) with Arc Max. What I love about them is the way they are using AI to provide value to users, they found good use cases to apply AI when actually browsing the web.
-
OpenAI had their own drama leading to conspiracy theories around finding AGI internally. For me, it was more a political move, but it gave me enough context to understand better things outside the Google bubble. Including a list of companies some OpenAI employees would prefer to go rather than Microsoft.
-
Some companies like Semplice (that I respect a lot), made an statement regarding avoiding using AI as part of their product. A statement that I totally agree with. Not all the companies or products should or need any random AI feature. We are still exploring frontiers, and it will take a while until we arrive into the right place.
-
I've got some interesting interactions with AI, first within my new Pixel 8 Pro (the first phone to get Gemini Nano). I was really impressed by Magic Editor, and how it is sooo close to what Adobe did with Generative Fill on Photoshop with even less power.
-
And then by trying Pi by Inflection, a type of AI designed to be kind and helpful, one that I used for extending my conversations with my therapist. It is as weird, as helpful as it sounds.
Some extra content worth nothing
Even before I changed my mind, I was already consuming content related to AI. Not only this content helped me understand better how the technology works, but also how to use it on my own day to day and by extension applying into the products I design for.
-
The Deepmind Podcast with Hanna Fry. A great podcast with just 2 seasons that talks about the history and achievements of Deepmind. It gets into the basics of AI.
-
Is AI the end of photography as we know it?. Becka for The Verge. A great recent video on how creating pictures through AI could be a form of art on itself.
-
People + AI Guidebook. A great source for designers to start when trying to get into designing AI for people.
Some thoughts around Conversational UI
After all this months of re-learning, I read again my own tweets. It turns out, I'm against chat bots, not AI per se. This is a larger conversation that I would like to address in another post but basically, for me conversational interfaces shouldn't be the only way to interact with this new type of machines. But it means we are still in the beginnings of exploring how to interact with LLMs and for now (even with multi-modal) prompting will remain as the main communication with them.
Taking Shortcuts
Last, but not least, the thing that finally drove me towards AI was my own company. When we started working on it, 4 months ago, I was already switching my thoughts, but actually working with AI and applying it, was the last thing I needed to believe in this massive and (now for me) exciting technology.
In some way, I'm still a skeptical, but definitely not in the same way.
"Are we really aware that moving fast with AI could become very dangerous?
Researchers have been warning about this for years and yet we like to see people like Sama go all on it.
On a business level it is great, but what about other consequences?"