Looking at the examples... Was someone using an LLM to generate a meeting agenda?
I hope ChatGPT would go berserk on them, so that we could have a conversation about how meetings are supposed to help the company make decisions and execute, and that it is important to put thought into them.
As much as school and big-corporate life push people to BS their way through the motions, I wonder why enterprises would tolerate LLM use in internal communications. That seems to be self-sabotaging.
You will machine generate the meeting agenda. My machine will read the meeting agenda, read your personal growth plan, read your VP's quarterly objectives, and tell me what you need in the meeting, and I will send an AI to attend the meeting to share the 20 minute version of my three bullet point response.
Knowing that this will happen, you do not attend your own meeting, and read the AI summary. We then call it a day and go out for drinks at 2pm.
True. Meanwhile, Sally in IT is still earnestly thinking 10x more than all stakeholders in her meetings combined, and is baffled why the company can't execute, almost as if no one else is actually doing their job.
You and I will receive routine paychecks, bonuses, and promos, but poor Sally's stress from a dysfunctional environment will knock decades off her healthy lifespan.
Before then, if the big-corp has gotten too hopeless, I suppose that the opportunistic thing to do would be to find the Sallys in the company, and co-found a startup with them.
Never. 100 years of unparalleled technological progress and productivity gains have lead to a society where 96.3% of the american labor pool is forced to work. Why should AI be any different than any of the "job saving" inventions that came before?
In the AI utopia, "knowledge work" is delegated to computers, and the humans who used to do productive and rewarding things will simply do bullshit jobs [0] instead.
Even already today, a lot of knowledge work jobs have a lot of overlap with the bullshit working with excel sort of office jobs especially when you consider what you are actually doing day to day and week to week.
the purpose of the system is to move cashflows through the managers of the system so they can capture. So no sufficiently large system can get rid of the humans it is designed to move money through unless there is some catastrophic watershed moment, like last year, where it becomes acceptable and an organizational imperative to shed managers. Remember, broadly the purpose of employees is to increase manager headcount so manager can get promoted to control larger cashflows.
No, seriously, there are rules having nothing to do with AI that require certain things to be done by separate individuals, implying that you need at least two humans.
Yeah. Almost everytime I see someone excitedly show me how they've used ChatGPT to automate some non-marketing writing I just come away thinking "congratulations on automating wasting everyone else's time". If your email can be summed up in a couple of sentences, maybe just paste that into the body and click send!
Yeah I can understand its use when it genuinely is in a context where presentation matters, but for internal, peer-level comms it feels like the equivalent of your colleague coming into the office and speaking to you with the fake overpoliteness and enthusiasm of a waiter in a restaurant. It's annoying at best and potentially makes them appear vapid and socially distant at best
Of course plenty of people make this mistake without AI, e.g. dressing up bad news in transparent "HR speak"/spin that can just make the audience feel irritated or even insulted
In many cases plain down-to-earth speech is a hell of a lot more appreciated than obvious fluff
But rather than being a negative nancy, perhaps I will trial using ChatGPT to help make my writing more simple and direct to understand
social norms are an evolved behavior, and especially necessary with people who are different than you (i.e. not your buddies who are all the same). ignore at your peril
AE is a special case. Procurement law for public agencies in the US requires qualifications-based selection for professional services. The price is then negotiated, but it's basically whatever the consultant says it is as long as they transparently report labor hours. This leads to the majority of effort being labor-intensive make-work pushed to expensive labor categories. There is no market process for discovering efficient service providers. This is part of the reason why workflows for transportation infrastructure design haven't improved in 30 years and probably won't until the legal landscape changes.
The instant I heard about chatgpt I thought one of its main uses would be internal reporting. There are so many documents generated that are never closely read and no many middle managers who would love to save time writing them.
Perhaps they asked for an agenda, so they can get a 'nice' example to mimic/use as a template (e.g. remember to write times and duration like this "09:15-09:45 (30 minutes)"
Or perhaps people are poo pooing a useful tool and they asked it something like "read these transcriptions from our many hour long workshops about this new project and write an agenda for a kick off meeting, summarise the points we've already decided and follow up with a list of outstanding questions".
Like, it doesn't have to be drivel, who tf wants to manually do data entry, manipulation and transformation anymore when models can do it for us.
Corporate bullshit is the perfect usecase for LLMs. Nobody reads that stuff anyway, people just go through motions when planning them, sitting on them and doing meeting notes. Just let AI do it! No need to even pretend.
I hope ChatGPT would go berserk on them, so that we could have a conversation about how meetings are supposed to help the company make decisions and execute, and that it is important to put thought into them.
As much as school and big-corporate life push people to BS their way through the motions, I wonder why enterprises would tolerate LLM use in internal communications. That seems to be self-sabotaging.