I just realized that I haven't been writing any content for a very long time on the Hive blockchain, and it's time to change that. Over the past months, I have been searching, exploring rabbit holes, and trying to figure out many different things around AI. To be honest, I have arrived at a stage where I feel a little more uncertain about the future than I did before, so it's tricky to define what the future will look like — but I think that's actually an important thing to sit with.
Automating the wrong things with AI
Everybody says "AI will change the future, AI will replace people," but I'm not so convinced anymore, because AI still has a lot of flaws. There are many ways AI genuinely helps, but there are also many areas where it simply doesn't fit or hasn't reached the necessary maturity yet. When I see people using agents to answer their emails, they are essentially delegating what I consider the most important thing: communicating with other humans. If machines are answering machines, what's the point — and how risky is it? You risk alienating your business partners and friends. How do you feel when you reach out to someone and get back a response clearly generated by AI?
The feeling is not different from reading a comment under your hive post and you realize it was written by a bot.
The workflow chaos
The whole world of agents and agentic workflows also looks like a big mess from where I'm standing. People are building pipelines and automations, yet few can clearly point to a business actually making money from it. Those who claim they do often seem to be painting reality in a rosier color than what's really happening.
When the numbers don't add up
After exploring AI image generation, AI video generation, vibe coding, and agentic workflows, I would say we are far from the point where AI will replace real people. I'm also noticing that many companies are quietly backtracking on their agent deployments — and it makes sense, because agents are energy- and token-hungry systems that can quickly spiral out of control. If running an agent costs more than hiring a person, and on top of that introduces mistakes a person wouldn't make, you have to seriously question whether it's worth it.
The right framing: leverage, not replacement
Even as models improve and efficiency increases, I don't think we are where the hype suggests we should be. AI is a great tool — but not one built to replace people. Where it truly shines is in amplifying your own capabilities: helping you implement something step by step, or teaching you the best way to do something you wanted to do anyway. That's the real value of AI. We should stop framing it as a replacement for people or workflows and start asking instead how we can use it to leverage our own potential.
With @ph1102, I'm running the @liotes project.
Please consider supporting our Witness nodes:
