Unless you are physically in the same room with someone, you can no longer know for certain that you are talking to a human being, rather than an AI proxy. An AI can do a pretty good job these days of imitating a person, at least for short conversations.
If an AI agent has been trained on years of someone’s emails and texts, it can create a plausible simulacrum of how that person speaks, and even what they might choose to say. This might eventually lead to some interesting situations.
For example, sometime in the future you might decide to take the day off, and let your trusty agent respond to annoying emails and texts in your stead. As AI improves, your little “vacations” might become progressively longer.
At some point, you might just decide to let your AI agent take over those annoying tasks entirely. Meanwhile, your colleagues may be having the same idea.
Eventually, all business correspondence might actually be conversations between AI agents.
I wonder whether anyone will notice.
I think the algorithm for much of business correspondence has become:
1. Sender has a a vague idea and writes a couple of bullet points.
2. Sender uses AI to auto-generate a wordy proposal, expending gigawatts in the process.
3. Receiver runs the wordy proposal through AI to condense it down to a few bullet points, expending gigawatts in the process.
It’s rather frightening to contemplate a future where that communication pattern becomes pervasive in all aspects of life: filing taxes, requesting medical care, applying for a job, asking for a raise, appealing the rejection of an academic paper.
Ah well, this is the future that we made for ourselves.