How are you feeling about AI at work these days?

​@LaurenClev 
It’s less of a race and more of a collective learning curve. I’m fully sold on the possibilities and have leaned into the 'energized' camp—it’s already become an indispensable daily partner for me, both professionally and personally!

I incorporate AI in what we do.   In Workfront, we are the leader in Form Fill-in (we helped scope it).  This helps us with Order Intake as well as creating shipping documents.   With Project Health, we utilize AI to summarize various topics based in the type of project we are looking at (Corrective And Preventative Actions, Building Maintenance, Health and Safety, etc)     AI is used to summarize customer orders, read documents, support negotiations, develop internal white papers.   We are working towards AI being a team member in what we do.   We are embracing and learning new ways to utilize AI!!!

5. Actively opposing due to its ecological, economical and socialogical destructive and exploitive nature. Will not adopt, will not use. 

You asked for a story from my day-to-day...

AI is new to me, and I think it’s very interesting that:

1. my company has sent me to an AI training course where one of the main takeaways (maybe I misinterpreted) is that AI is error prone and always needs to be validated
2. my coworkers are using it to summarize documents

I can’t figure out any way to validate a summary of something without reading the whole thing. To give you one example, we have document summaries of 12 hours of meetings… and I’m trying to figure out how to put into practice what training is asking -- are they saying I still need to listen to the 12 hours and then validate that the summaries are correct? Or what is the threshold for validation where AI is concerned?

sorry, Story 2 -- I feel like I had to ask myself a similar question during the Champions Forum in Nov, when we got the opportunity to test a few new AI tools. For one of them, the request queue file upload to fill out the form (the feature Scott was talking about), a few of us were confused at the answers that AI was coming up with, as they didn’t appear to be coming from the file upload. If we move into a future where we get to upload a brief in order to fill out a form, and the functionality comes up short, I kind of wonder if anyone’s going to notice (I know we were all more focussed in other directions during that particular Lab, so this ended up being a followup question I sent to the product team.)

​@LaurenClev Interesting poll question.  However, in my honest opinion there are missing poll answers, as in “Cautiously optimistic, but also realistically skeptical.”

I would hope by now that people understand that “Artificial Intelligence” is, at best, a misnomer.  I have said in many discussions that someone Marketing individual along the line decided that calling it advanced automation against a Large Language Model database would not sound nearly as exciting.  Case in point: just look at how far we got with Hadoop. All people could remember was an obscure picture of an elephant and... maybe a hoola hoop?

Nope! Not this time, they decided AI was a lot easier to “explain” and perpetuate than LLM.  Never mind that even though it feels like you’re talking to an adult when you submit your prompt into the computer, what you might get - if you’re lucky - is the amassed data and response of something that might resemble that just short of a three year-old’d intelligence, and even then I’m being extremely generous.

That’s because the automation (the Wizard behind the green curtain) driving all that logic is nothing more than an algorithm doing its best to predict what it thinks you may want it to say, based on all of the possible data it’s been “trained” on in the past.  The only problem with that is the information it provides is often wrong, and then you spend more time trying to convince the AI why it’s wrong, while it keeps admitting it’s wrong, while it justifies why it was wrong, while it keeps regurgitating more wrong answers, and then continues to produce even worse results the deeper down the rabbit hole you go.

Of course, magic always sounds exciting at the outset, but that’s because people don’t understand everything involved in the contract.  As soon as you dive into the messy details of what’s required when it comes down to what you just agreed to, all of a sudden, that request for a “magical result” isn’t quite as palatable as you thought it might have been.  Just like all of the classic Disney movies always say, “Magic always comes with a cost.”

So, my official answer right now is “Overwhelmed,” because I honestly believe most companies are diving into AI with very little understanding of what rabbit hole they just jumped into, and unfortunately, there are a LOT of Alices chasing way too many AI rabbits.  On the flip side of that, I can say I am optimistic in that there are people out there who do understand AI can be used as a tool or framework to use as a STARTING POINT.  And that’s where I will stop, because that’s where the conversation begins.

How many other users are using AI as their starting point and not as their finished output?