As AI gets shoved into everything, it’s time to think about its longer term impact on computer interaction.
Not too long ago Chrome has suggested to use its “AI” mode instead of searching. I have tried it recently, but I’m not really a fan (as with anything AI related at the moment). There are two reasons:
- It’s the same low quality output as the AI slop output that normal Google AI “answer” gives you. Answers are indirect and it would rather please you than tell the truth. (Also see: You’re absolutely right!).
- The time to generate responses is slower than reading search summaries. It doesn’t shut up easily which makes the results more bloated and also making it use up more energy, water and RAM.
Some will say that if you prompt it correctly, it will magically come up with whatever you were asking for. I think in a lot of cases it’s on par with search results and sometimes better, but that’s only if you ignore the cases where you need a very specific answer as opposed to one shoved through a GPT. Prompt engineering can only go so far, given the “ceiling” of an LLMs output will always be less than what’s available to it as training data (a bit of an educated guess, but it makes logical sense?).
One day search engines will no longer be the default, instead Chrome’s Omnibox (address bar) will be an AI prompt field. Most people won’t willingly ask for it, or would even enable it as an opt-in feature, but all it takes is one update for it to become the norm.
Technology has been integrated into almost everything for so long now. Mass data extraction and analysis, with or without the user’s consent*, has been operated by Big Bad Tech for long enough, yet alarms aren’t ringing in peoples’ heads. At least in the UK, where we seem to be OK with being friends with America at the moment, compared to Europe who seem to do a better job of abandoning for-shareholder services/products, which at this point only really exist to make life hell for the rest of us.
But I fear that not many people outside of our technology nerd circle will care. Despite the layman not asking for it and “AI might take my job” going around for ages, it still became the most downloaded app by far.
When generative AI first broke the internet, it still felt very experimental. Bard, Duet AI and Bing Chat are all in the past. We aren’t at the beginning of the race anymore, laughing at Google telling people to eat rocks. Remember how crap the image generation was, circa DALE-E mini in 2022? Now a lot people are easily fooled by it. I can go into town and buy t-shirts with AI slop on them. It’s here to stay in some sort of capacity.
[rant]
So far this post make me feel like I’m part of Just Stop Oil or XR (remember them?), announcing there’s some sort of threat which to the rest of the world, feels tangible, but a bit abstract at the same time. We can see these thing are “bad”, but they are beyond our capacity to care. I kind of which it wasn’t this way, but maybe it will have to be, for better or worse.
[\rant]
I don’t think this is inevitable at least anytime soon, but if we are going to live in a world where machine-generated content is the norm, how can those who don’t put up with it?
The “hacker culture” that people associate with the old days (e.g. early Linux days) exists in one way or another. Sure, everyone has computers in their pockets now, but I can still categorise people based on if they want to hear about my self hosted services.
Will I exist in a bubble made up of those people or still be able to integrate with the rest? I don’t know the answer to this, of if I will ever have to make this decision. Time can only tell. It’s also hard to judge how we could have an influence over governmental decision making with the next big SaaS/AI thing. As I said, other states are better than others, but at the rate Trump 2 is going, it might be hard to avoid or undo.
*Legally, clicking the button that’s a different shade might be consent, but did you really know what you agreed to?