how many times will LC say AI. Drinking game!
53 replies (most recent on top)
@xq the biggest challenge is incompetent leaders who demand solutions but do not understand the most basic concepts required to deliver on their cockamamie vision and think “AI” can just make it happen.
@w5 that’s cool bruh
while you get to attend Ignite or databricks or Gartner or whatever waste of money conference that allows you to travel, the AI team gets to stay in place and work their as--s off in the office. That’s another clue for anybody actually working in this area - everybody knows EAI is all work, no play.
What is funny is that anybody who is actually in IT and attended the IT Townhall, including LC and his LT, have no idea what everyone’s talking about here. Because IT is irrelevant.
@w1 you’re right in that that work was at least from 7 years ago. It just took that long for Chevron to decide to communicate it externally. The entire (what’s left of the) data science group worked on that thing a decade ago. And the powers that be decided to highlight it in a recent investor presentation, calling it AI.
Sounds like you haven’t been paying attention internally or you’re a troll from the outside to think that it’s the only things chevron has done in “AI”, based on what you saw in the external communications. It just takes that long until PGPA feels comfortable talking about anything we do. Like talking about sh!t from 7 years ago that everyone already can do instead of the actual good stuff.
Making pipelines out of well reports and chatting with that data has already been shipped to India. Chevron isn’t focused on doing that type of work. That’s at least 2 years ago when they started doing it.
@xq Chevron’s doing those things you describe. And guess what? It doesn’t matter as much as you think it matters because your strategy and business development groups making decisions don’t care what you say you can do.
The whole industry is full of dinosaurs making the actual decisions. You think you’re cool, but you’re not. The biggest problems in energy are not what you think it is.
What’s your biggest challenge?
Are you focused more on genai hype or are you focused on delivering value which is probably in far more traditional ML models?
Where I worked we fully automated sub surface detection of salt domes and other traps end to end with zero humans involved. By the time we finished, we didn’t have or need a subsurface department anymore. We had a setup where we could slap in raw seismic data and just sp-t out conclusions and confidence intervals.
As far as I can tell the everyone else who is small to large is extremely attracted to making pipelines out of well reports and adding AI chat to random things instead of asking how to use AI to solve the biggest problems or cut the most expensive items on the bottom line - and what types of AI really matter for doing that.
Most energy LT is just clueless and says stuff like “ask chatgpt and copilot everything” with no further insight. Everyone thinks they’re at the top because nobody smart who is still working at the top is sharing anything that has been done so it looks like a level playing field of awkward shuffling.
Well, if you can’t do the things I have described and aren’t trying you’re behind. So happy I left this industry. It has such a crab bucket mentality. Big tech life isn't easier but it’s far more rewarding in every metric.
I think this is an industry wide problem. Anyone who can deploy an AI system start to finish has no reason to remain at an oil and gas company where you throw away your time on politics and access issues (logins, data, installations, etc). Especially when it offers increasingly less pay and benefits than numerous other options and far, far less job security.
BCG and McKinsey won’t fix anything. How can they? They’ll have the same impediments and even less domain expertise. If anything they’ll make it worse by accelerating talent cuts and dropping bo-b projects that don’t scale.
What’s going to happen is anyone smart with AI skills will leave the industry entirely for greener pastures and easier days. Not easier in terms of workload or goals mine you, but easier in the sense that your time is all spent visibly moving the needle instead of moving old fogies and begging for data. More time on rewarding challenges. I don’t mean to imply at all that you can coast somewhere as an AI engineer.
Some kind of mid size specialist firm will rise to fill this niche with SaaS and no major will have in house talent outside of a high turnover hub in India. Maybe they get bought by schlumberger or halliburton, maybe not.
As an energy AI engineer myself, the thing I now work on is using cheaper models and shorter prompts to save tokens. If your energy AI org has not deployed so many useful things that they are now focusing on cost optimizing them at the token level, then they are not at the top. Odds are high that tokens will get a lot more expensive as the venture capital that allows cheap token price wars dries up and other even harsher bottlenecks emerge, like data centers and electricity for them and even fast ram. Do you have a plan for when LLM calls need to be rationed?
In 3 years you’ll look at this post thinking I had a crystal ball. Save it.
I am a data engineer who left Chevron for another industry, and was involved with several data science or loosely AI projects, but I was connected with many working in data and AI still
You are not behind. Your issue is that Chevron is too damn big and there aren’t enough technical people who know wtf they’re doing to cover all the areas you need to cover.
And to the person who said chevron is only full of consultants doing AI, that cannot be farther from the truth. Your data leaders are quite picky and maybe have too much hubris to involve consultants on their work. I tried to get in the group in the last, but was not able to get in for who knows what reason. More people will leave, and you’ll be forced to use consultants. That time is coming unless LC decides to do something to stop the bleeding which I know he won’t. And that is why I’m gone.
But the grass isn’t always greener on the other side. It’s harder work but the pay is good if you’re just a few years outta school.
@x9 yea im really wondering what conference these guys are talking about because there were several in the last couple of weeks. The guys (same or different?) talking about flat files and only doing decline curves probably aren’t involved with the actual AI work at Chevron because when I compare notes with people from other operators, Chevron is definitely not behind.
And yea we need to do a better job of letting random people speak at conferences and misrepresenting data science and AI work.
@w5 I’m in this space and have talked with technical folks actually working on AI projects at several other Oil and Gas companies and can confidently tell you that Chevron is not behind our peers - at all. We’re all struggling with the same challenges while everyone thinks the others have figured it out.
I met 50 people. Have any of you ever been to a conference? Do you note everybody who grabs a mic, add them all on linkedin, then scrutinize their profiles, then interview their teams to see if they’re “legit”?
Do you even realize how ridiculous the expectations in the comments are? I’m just a person from another major who is sharing what was shared by someone claiming to be one of you high up in AI LT.
If you don’t like it, tell some authority figure to lock down conference attendees because it’s your companies job to be the id--t speaker police, not mine.
Some friendly advice - you need to have centralized data access in a huge data lake with ways for people to request a connection. You need company wide designations for confidentiality and a clear chart showing how much risk is involved in sharing document information for each level. You need access to an AI API available to developers in
order to catch up to your piers. If you don’t have that, then you can’t replicate and scale AI solutions. You are doomed to make infinite point solutions that only work for one little piece of one team via manually upload or consuming a sharepoint or network drive. You certainty can’t do something like train models on data from multiple similar facilities or pieces of equipment located around the globe.
You are not really AI capable without empowering user data access and maintaining data quality. Two types of companies were in the room, and Chevron sounded like the no access, nothing built, throw money at consultants kind based on your self proclaimed AI speakers. Your mileage may vary, I don’t work there.
@a6 100 percent we are so far behind our peers on AI. The most useful one theyve made it advanced mass decline fitting which you could get in software packages 7 years ago
@vn You gotta ask what's making senior IT leaders walk. What do they know that we don’t? The next few years look rough with that oil price plummeting. It's not just the onshore folks who will feel the pain; the whole company's in for a bumpy ride.
What IT Townhall?
Did anybody actually attend?
Just hearing many people leaving the company. Latest one being a senior IT leader. Even they are giving up.
@sm hmmmmmm I went to a conference, too. And it was quite a different message than the one you described. Maybe go to a more reputable conference with real Chevron people?
@r8 cool story bro
@sm name names. What conference was this? You know every message is sanitized by corporate affairs… and we have never heard that.
sure maybe. he actually
said this on a microphone to the whole room which made it really funny
i hope you understand that i don’t roam around conferences asking everyone to verify their employment and show id and then proceed to rank their legitimacy amongst their peers
Wow my bad the autocorrect was bad lol oops. U get the point.
@r8 I can fake news on that one. What US AI leader would actually spew that nonsense. Unless you’re taking to someone with a fake title.
I met some US Chevron AI leadership at a conference just last week and they said that they won’t do anything unless someone else does first and then they’ll just copy it.
Innovation level zero.
They also said no regular employee can get any kind of data unless it’s flat files from their own business. Thats even worse. AI at any place is impossible under those circumstances.
It is truly embarrassing how the culture of this entire industry is to cut costs, see any investment as an expense, and not do anything unless somebody else does like a giant fu--ing chicken.
Moonshots is a silly phrase to use
@ca you mean bonus airy
@ar if it’s the guy I’m thinking, he is annoying AF
The lady from Buenos Aires saying that their employees WON'T COME TO THE OFFICE 4 DAYS A WEEK WAS BASED!...
LC's idea of AI is to use ChatGPT. Ask it anything and you get all these great answers. Maybe we can ask AI what does MW or MN do all day.
Prediction:
ENGINE will claim to be doing AI and be stuck in POC with no use.
IT employees in the US will be scrum mastering in ADO purgatory until they just quit themselves.
There’s no budget for high cost contractors here in IT… so that’s not happening.
Prediction: IT will contract with high priced consultants to come up with AI ideas. Chevron has NO capacity or capability to do this themselves. It’s sad. Then we will fail miserably at implementation and maintaining and recovering the business case. Broken record…what we do all the time, miserably painful to see.
@a7 1960 Gov't & NASA....Lordy...There was leadership there!!!! Here, perhaps not so much....
We are cooked as a function
A whole IT townhall about AI, and the AI group got taken from LC and put into another org.
You can’t spell India without AI!
@aq whoever the guy that asked that question was is a bad a-s, at least someone has ba--s.
The question on why other locations can’t work 2 days hybrid is getting upvoted but they didn’t answer.
Forget about that KT now
@ak this wasn’t a town hall, this was just an AI glaze job
That was the biggest joke of a “town hall” they ever done. It should have just been called “Our AI Future” and been a news article. Absolutely nothing to address any material concerns the rank and file have about actually working and doing their jobs. A new low.
Wow. BA is 2 days in office as HOU is going to 4 days. Shows where they want to retain and where they’re willing to let go. No future here
Only US has to resume 4 days in the office. Other tech sites will remain hybrid.
How will IT handle everyday in office when there aren’t enough desks for IT?
The number of people in the room speaks volumes. Lots of empty chairs.