Google Meet’s new AI will be able to go to meetings for you
theverge.com
theverge.com
- write emails
- go to meetings
- any other type of activity that involves sharing information.
Humanity didn't develop emails, or have meetings, or communications because it some sort of ritual to a strange god (barring the meetings produced by the priests of agile to appease the Agile God). We communicate in order to share information from one person to another for a reason. So if we have a workflow where someone uses AI to generate an email, that results in a meeting that is then summarized by AI, that is then sent to an inbox that is then scanned and the contents of which is summarized by AI. What was the point of any of it?
At no point did a human learn any new information. It was all just various LLMs generating symbols that changed nothing in the real world.
What is the end goal of this? Or is the hope that if we get enough AI hooked up to each other it will magically just start creating and selling products and money will magically appear and everyone will be happy?
At the end of the day the internet revolution was all about Information Technology. That information exists to be consumed, understood and acted upon by human beings, to change our understanding of the world. If we are just having AI talk to AI to talk to AI, it was all an exercise in pointlessness.
I think what this tends to reveal is just how much of a house of cards the modern business world is.
There is a basic conflict then between business (who optimize for cost) and government (who optimize for stability).
And it's systemic. The product of management is reports. The product of management is meetings. To them (and the people who ultimately run the company) these are the solutions. Have a meeting, write a report, do a review. Those are the tools they know. They have hammers. Everything is nails.
We just went through a season of layoffs - I find it hard to believe that no one thought to get rid of the so-called bullshit jobs for greater efficiencies.
For arguments sake, lets say the conspiracy for bs jobs goes all the way to the top (C-suite, boards) in established companies: how about startups then? Where are the streamlined startups that are more efficient than incumbents by simply not creating those "unnecessary" roles?
I suspect that coordination problems necessitates these roles as companies grow larger - call it corporate carcinization if you will. If you were to start your streamlined org today, it will evolve into a crab-like form.
The business model of a startup is to work everyone you have beyond their natural limits, this doesn't scale to an entire economy.
I know a few startups that laid off most of the revenue drivers in sales and marketing, but kept expensive founders and coders. They are on paper leaner, but not necessarily more efficient.
The problem is that we often think that if we don't understand something, it must not be valuable; and I think this problem is exacerbated by high levels of education and/or intelligence, because we usually understand things, which must mean the things we don't intuitively understand are obviously not valuable.
The reality is, there is some bloat which gets introduced as businesses grow. But it's not as much as we often think.
What blows my mind is how much engineering effort we collectively put into consumerism and business process, without taking a step back to the design phase and asking ourselves if we need a new architecture.
We do however have lots and lots of ritual as to how those communications occur in order to confer the proper respect and tone. In most cases its pretty much noise but you can't skip it because skipping it means offense was intended. We can all recognize this but we aren't really free to escape it.
Generating emails with AI lets a person avoid the appearance of intentional offense and summarizing emails with AI lets you cut to the point while acknowledging that you are dispensing with all of the ritual.
Who knows, maybe after long enough the cultural convention can shift to just sending your prompt straight to the other person!
Obviously I'm skipping over a ton of pitfalls but I think that captures the appeal for a lot of people.
This paper used 3.5, not even 4 and is basically a couple AI agents collaborating to create software.
https://arxiv.org/abs/2307.07924
This is another example at multi AI agent collaboration.
https://arxiv.org/abs/2307.02485
Getting relevant intelligent text to arbitrary previous text means generating text capable of making decisions. That is the end game here. Your machine can think for you now.
Because we didn't have a machine that can do this whole collective decision-making process. What if we make an AI that's powerful enough to replace a whole organization? A whole company? A whole industry?
I won't expect the robot manager to have a robot secretary, who complains to the RR department when manager makes inappropriate remarks.
He also mentioned writing emails with ChatGPT (though I can't remember if these were for communicating with the devs, but it doesn't really matter). I remember thinking that well, you my friend are pretty ignorant and not someone I'd like to work with. Even unethical.
So someone puts in the work to inform you about important details of their work/thought process and then this guy just ignores the whole thing. Even if the developer in question is really sharing more details than needed the right thing to do is to actually manage them and teach them to share less and make their own decision. (Hard to tell from the outside, of course, and the manager in question might be wrong too.)
Now if someone also blows up their own emails with generated details, especially if it's toward the same person whose email they have compressed with the AI then besides robbing the time of others, they may successfully deceive others and themselves and pretend that they have the knowledge what they don't have. And that will hurt the decisions made by them and by others. Basically the shared reality that you want to create with communication will not be a shared reality but out-of-sync views and you'll end up with a lot more "oh, but when I said that I thought you meant X...."
The worst-case scenario is that human communication becomes completely routed through this LLM-in-the-middle.
Shareholders will be happy the day a company produces a new product and no human ever worked on its development.
All of the profits, none of those pesky workers demanding silly things such as payments and OSHA protections.
Yeah, you say this like it's obvious, but it's quite an honest and productive line of inquiry.
But whatever the history of meetings are, it's undeniable that a lot of the current ones are merely ritualistic.
And yes, the one single way to deal with the ritualistic meetings is to not attend them. I am nearly convinced that this headline is an attempt of modern art, instead of objective information. Even though the application looks like something with real goals.
We were paid to make and use those technologies because we were taught that’s how society works.
Fiat economics is not solving human problems. It’s propagating fiat economics.
Humans have no divine purpose. The universe does not force social norms on us. We do that to each other.
You exist to pass the butt… dollars around. Fiat economics has filled the biological hole religion stumbled upon, probably helped evolve.
The only reason for any of this is deference to those who came before.
If one day everyone sends an AI to attend a meeting, will all these AI's (which are all instances of the SAME platform!) still have to act out a human meeting...?
Reminds me of the "Electric Monk" from Douglas Adams' Dirk Gently novels, which is an appliance whose job it is to believe in things on the behalf of its owner...
I think this is the absurd part. I can see an obvious use for some meetings, for example: where lots of people are invited 'just in case'. You could get the AI generated summary and see if anything relevant came up, then follow up if it did.
I feel like this says more about a somewhat bad meeting culture that a lot of us partake in and less about the inevitable rise of AI.
I get the value in being in on a meeting is popping in when necessary (ask question, clarify something from your team, etc…).
Many meetings could also just be done more efficiently as chat discussions.
To me what this says is that devs like having their thinking done for them and the requirements spoon fed, instead of having to actually make decisions.
Although looked at that way, this is really Google's entry into the CI space.
But, like, actually delegate. Let others make decisions. Trust that they know more about an area than you do. Just because you've read their notes doesn't mean you're an expert.
Trust your team.
If you don't feel comfortable enough to say, "This could be an email thread instead." or "Is this meeting really necessary?", this is probably a great fit for your org, who will eat this up as "innovation."
But here’s a more science fiction spin on it:
Instead of thinking of this as “going to a meeting” think of it as “representing you.” That is, maybe it can try to answer questions on your behalf and it can give you a summary of what happened afterwards. Everyone will know it’s a representative, it can’t commit to anything, and it might not be entirely accurate, but maybe it’s better than not being able to ask you questions at all? (For example, when someone’s on vacation.)
I don’t expect that sending a person to represent your interests is common outside of maybe upper management and of course the legal profession, but perhaps it’s useful. It might be better than sitting through an entire meeting just in case someone might want to ask you a question?
I wonder, though, whether it would be better to just make a chatbot available for people to talk to at any time without it being a formal meeting. Also, what information does your representative actually use to answer questions?
Maybe a chatbot could be a way for people to make their semi-organized work notes available to others? Or make a vacation auto-responder that’s trained on a subset of your outgoing email.
Imagine what happens after someone leaves. “Sorry, she’s not available, but you can chat with her bot.”
We won't even need a calendar then to schedule it, you just write some text and press a button to have a machine agree/disagree with you.
Soon people will wonder why this process is even called a "Meeting"...
I can almost hear the AI-bubble inflating already...
A 20-year-old truck driver will go to trade school and become a mechanic.
As much as we like to think of ourselves as special, this has all happened before and will all happen again.
One out of three jobs lost in a single lifetime.
During the same time, the standard of living skyrocketed.
For example, meeting notes are notes all participants agree on (action items, owners, dates, etc.).
While other humans attend this will still be very compute-intensive, but as soon as all the attendees are AI's it will only take a split-second to generate meeting-notes straight from the Agenda! ;)
These are big-audience meetings usually with one person presenting and then a Q&A.
I read the summary and then if it looks interesting I listen to the meeting when it’s convenient.
(Not gonna shill for the app: it’s just one of the ones in the Zoom “store” — there are several).
I think we humans do just fine when we can't attend meetings. Most are unneccessary anyway.