notes from a tech conference

[screenshot of direct Teams message from my manager's manager's manager, asking me if i'm attending a conference tomorrow]

i'm getting forced into this conference, aren't i

8AM to 4PM kill me

i have boycuddles scheduled

wear shorts

i show up in pup gear. what are the gonna do, boot me out

(probably)

that would be so based do it

i will live tweet the con

they have a piolted boston dynamics dog going about

we start the conference with a pledge of allegiance and a prayer, in that order

they have a projector screen up for the conference stage, sure, fine. but for the pledge of allegiance they run a looping video of an american flag waving, and then for the prayer they run sun_rays_through_clouds.mp4

"now we don't have flying cars yet -- Elon Musk is gonna take care of that"

"remember Star Trek? those communicators? they're in your pocket now"

it's iphone 2 guys iphone 2 real remember iphone this is iphone 2

the conference tagline is 'Production 4.0: Streamlining for Outcomes"

it's iphone 4.0.guys

very noticable amount of white noise from these speakers

[my regional manager]: "innovation, from a graph perspective, is accelerating"

opening keynote is [Bob Smith], cofounder of [EngageStart], a "full stack AI advisory company". talk is "2025 for AI: Turning Theory into Practice"

BS: "show of hands: who owns crypto?"

BS: "any etherium in the audience? Bitcoin holders?"

1

???

I'm going to throw up

I'm so sorry you are experiencing this

the things i do to remain employeed

do you feel like a real innovator

they handed out joke million dollar bills and we're playing a weird stage "investment" game

BS: "the average IQ in this room is pretty high"

this is what happens when you and all of your friends are MBAs

he got five volunteers to read out five blurbs from five fake companies. four of which are coded as "legacy", "establishment" companies and one of which is coded as a "tech startup"

and we "invested" our comedy dollars by vote, and then he revealed that the tech startup is actually Ford Motor Company in 1903

BS: "we're not in 1903 though. in 2023 they launched ChatGPT"

there's a joke here about nazi sympathizers sticking together

BS: "with AI coming... where are your personal ruts [hangups]?" with a really stretched car metaphor, like a rut in the road

BS: "either you're disrupting your career or someone else is. this was a fun thing to say a couple years ago" you've been saying this for YEARS bro

1

BS: "as you go on your AI and ML journey, you are going to skin your knee. and i want you to skin your knee"

These are insufferable individuals

[screenshot of slide, showing "Disruption is Everywhere", with a bunch of buzzword companies (Netflix, Uber, AirBNB, Amazon)]

Just put the fries in the bag

[screenshot of slide, showing that AI-generated Coca-Cola Christmas commercial]

1

Honestly

Yeah

They should be serving me canes

[screenshot of same slide, with headline "Coca Cola AI-Generated Holiday Ad Falls Flat" overlayed]

Ebu conferences are so draining. Sending my condolences

1

BS: "that headline is the noise. the signal is the the fact that they generated that commercial at [numbers that imply it was a fraction of the cost]"

oh gods he's quoting Zuckerberg

BS: "the three legged stool of AI: LLMs, GPUs, data and content"

BS: "Grok 3 came out this week" (Grok is Elon's twitter baby ai)

BS: "who in the room does not have a ChatGPT account? there's people who can help you after create an account"

BS: "Intel will be out of the business by the end of the year"

BS: "without your data these things don't work. and let me tell you a secret: we've hit a data wall. so companies are looking for data."

BS: "your phones and computers run on CPUs. game consoles and AI run on GPUs"

Uhm

Wut

BS: "if it's getting harder to hire... shouldn't we be pulling those responsibilities [into AI]? can't we do these plants with half the people?"

Kill

Rage

Anger

BS shows example of [retail company]. BS's company "cloned his [the CEO's] personage" in video and text format and "connected it to the company master calendar"

so if you hate working at [our company], imagine getting sent [an] AI [version of our CEO] every day for every event

BS is openly suggesting [my regional manager] AI clone himself and send videos to employees :)

BS: "aI iS gOnNa tAkE jObS!! No, it's gonna remove the tedium from those jobs."

this comes not ten minutes after he suggests replacing some jobs with AI

me <3 still have never used it !

BS's second case study is just... aggregating reports into bigger reports

reading this is infuriating

I hate tech why did I go into a career where I have to be surrounded by tech bros at all times

BS's third case study is an "AI-enabled" API connection to a payroll system

i'm not even sure what this is. this doesn't even sound like AI, it just sounds like previously they had a person do a lot of spreadsheet math and replaced it with a python script

also they didn't proofread their comedy dollar so it still has the text "this note serves to commemorate Barack Obama the 44th president of the united states"

1

hell yeah

BS: "your company needs to have a thesis [on AI] which is an opinion presented to the market"

smells like "buy now buy now buy now" but b2b flavored

BS: "i will send you our 29 page guide on how to use ChatGPT, and a PowerPoint that makes you a Prompt Expert"

QR code links to his LinkedIn

I bet he gets so much pussy

Gigachad

BS: "okay the question was what are the things AI is not yet good at"
rando in the audience: "how many letters in the word strawberry"
BS: "i would say creativity. we're not at AGI yet"

editors note: "AGI" is a conspiracy theory that "AI" (read: spicy autocomplete) will eventually take off and create superhuman intelligence

oh god he's a 10Xer

Wut

That

Thats hilarious

editor's note: "10X" is a conspiracy theory that some kinds of people, specifically software engineers, are capable with the right habits of consistently being ten times as good as their coworkers in terms of whatever output. usually coincides with an extreme superiority complex

Oh hes a narcissist

Recall the chiefmost of the commandments of the Orange Catholic Bible: "Thou shalt not make a machine in the likeness of a human mind." Have we so soon forgotten the wisdom of our elders?

The thinking machine is an Abomination

question from the audience: "what about AI accuracy? I've had yahoo emails summarized completely incorrectly and i don't trust it"
BS: "what's the over under on the human equivalent's accuracy? do they break up with their girfriends? do they leave for another company?"

1

Wow. Impactful stuff

His wife def just left him

question: "what's the real threat from China?"
BS: "There was a company, DeepSeek, that released a model on a Monday [ed: false, it was a Saturday] that caused a 300bn loss in Nvidia [ed: misleading, that's market cap, a made up number anyway]. turns out they used distillation, basically copied ChatGPT off of OpenAI [ed: as opposed to OpenAI that just happened to stumble upon massive scrapes of people's data in a box in their grandpa's attic]"

BS: "the Chinese are all in. they have cities dedicated to AI"

Oh woah! Racism too! Man

He just keeps getting better

I wish i could suck his fucking cock

God

BS: "the GDPR -- i don't need some 77 year old senator dictating my AI strategy"

1

oh he HATES the GDPR

if i mention the CCPA he might explode

i mean he's an AI bro. of course he's anti-regulation

this was in response to a question that brought up crimes (fraud, mostly) done with AI

NO THEY WILL NOT BRO. MORON

EXACTLY, YOU’VE HIT THE BOTTOM OF THE BARREL AND YOUR POSITION IS THAT THERE WILL TOTALLY BE MORE MONEY IF WE KEEO DIGGING THROUGH THE BOTTOM.

this is so funny i’m not even gonna quip, just acknowledge it

next speaker: [Paul Phillips], VP of [ProtectSafe], a company that bridges the gap between private companies and the FBI for the US gov

correction: not a company. an organization? NGO?

PP's talk: "Production 4.0: You Need a vCISO"

PP's talk seems to be a straightforward cybersecurity sales talk. basically hire a cybersecurity guy. which i mean sure

PP: "put forward a entry level cybersecurity job. got a PHD in business administration apply. told him upfront that he was overqualified and wasn't going to hire. asked him why. said that he learned that cybersecurity is a business problem so he wanted to learn all the business first. really opened my eyes. still didn't hire him"

1

PP just cited the Google AI summary for the "first CISO". this entire slide is AI generation

[image of AI-generated Google search results for "the first CISO"

PP: "there's a new thing called 'shadow AI' where your employees don't like your AI system so they bring their own devices and use their own AIs"

I hate ai so much

next talk. [Barney Manson], developer for [major agricultural industry company]

talk: "Deriving Insight with AI: A Comprehensive Approach to Equipment Failure Classification"

this shit is insufferable and i’m not even there

thank you for your service ebu

BM: "had a lot of bad data about failed pumps. basically we started putting it into our ChatGPT and getting back analysis"

[slide of diagram of Microsoft cloud service spaghetti]

BM: "anyone want to define what a prompt is? ... right, a series of instructions. what if you could save that prompt and use it again"

bro. the autocomplete does not understand you. it's autocomplete. it cannot follow instructions because it cannot understand them

[slide of them prompting an AI to analyze agricultural equipment failures]

oh gods. they're parsing JSON from microsoft-flavored chatgpt

editor's note: JSON is a text-based format for sending data over the Internet. there's no guarantee that any LLM will output valid JSON, or even JSON at all, no matter how politely you ask it

this is because LLMs are stochastic and random inherently at their nature and they blindly predict next-characters and next-words

oh. oh fuck. they're asking the word predictor how confident it is that its predicted words are true

[slide of them prompting an AI to output JSON with a made-up confidence score]

they're parsing these outputs into graphs and then directing company resources to the top problems. as "identified" by the word predictor

question: "are you using sensors to detect failures? preventative maintenance?"
BM: "yeah, we can concatenate that data and pass it as a string"

they're piping all this into SharePoint. they're using SharePoint as a database

BM: "every time we get results back and they're low confidence, we put the original prompt and the output as an example back into ChatGPT and ask it to generate a new prompt. we're currently on our 8th iteration of our prompt."

next talk. [Alexander Simons], robotics specialist at [major oil and gas company]. talk is "Advanced Mobile Robotics Powered by Autonomy and AI"

AS: "we originally started with just drones with cameras, and moved on to putting LIDAR and other sensors on them. [shows footage of drones and Boston Dynamics dogs with sensors on them]"

AS showing footage of drone operators going into hazardous areas of plants to do inspections and whatnot

also showing using LIDAR + cameras to do photogrammetry to build 3D models of plants. evidently they were previously using the 3D view in Google maps

it seems the sales pitch here is drone photogrammetry and then autosplitting the scanned model into parts for different jobs/people... but photogrammetry/LIDAR is rather noisy, especially on shiny surfaces (which: oil refinery plants are made of). might be more noise than signal

plus... don't you already have schematics and 3D models of plants? do you just built a plant and then throw away the build plan?

also [my sub-regional manager] came up to me and made a very pointed remark about not seeing me around... and then grabbed the uncollected name tags off the table presumably so she could know who didn't show up

1

oh in the Q/A he said it's not as good as human scanning. but they're going to "merge" their drone scanning and human scanning

AS: "none of these devices are intrinsically safe [ed: technical term meaning they won't interfere with machinery electromagnetically]" okay so nobody can use these

[photo of framed conference screens showing the name of the conference, but a modal pop-up saying "System UI isn't responding" is blocking the text]

bruh

next talk is two people. [Prasert Iamotat] and [Spencer Somerton], Augmented-Reality Platform Architect and AR-XR Strategy Coordinator respectively at [major agricultural company]

talk is "Intro to Digital Twins: Building Digital Platforms"

appears "digital twin" refers to objects, not people

okay in terms of meaningless terms per minute this is way up there

PI: "what is the digital reality ecosystem? [major agricultural company] is global, we aren't gonna have one digital twin tool. the common platform is data. so we need to have a digital system of digital twins where data is the common denominator. so visual data and geospatial data. the data has to be open, agnostic, and adhere to visual standards. there's multiple layers of data"

SS: "one of the key principles behind this whole thing is open interoperable digital twin. companies are looking for holistic digital twin for use cases. we're trying to break this into components of information, so it's not one tool, but split into five different pillars of information, democratizing information and data"

PI: "historically visual data was just visual data, but we need to have guardrails. and once data is open and interoperable, there we go"

wh

what

what are they fucking talking about

SS: "being able to lay out plot plans, or blast zones, or water maps, instead of engineers having to look in fifteen different locations, they can look in one location and it's all there"

word soup, i’m having an aneurysm

flooding the zone with shit

PI: "project stage for example. we have a 3D CAD model. take a picture. overlay it in the model. color code it. red means bad. green means good"

there is so fucking much i literally do not have time to type

PI: "imagine if you could see those work tasks visually. enhance visual inspections. we used to send people to do inspections. take photos. send a drone up there. take it one step further. train an AI to see anomalies. instead of folders and folders of photos. see those yellow dots? those are camera locations. think about geospatial positions, global positions, 3D positions"

SS: "being able to connect our workers in the field in spatial ways. give them the right information at the right time. and that gives us the right reality"

PI: "visual fidelity and digital accuracy is really what's gonna drive it. we try to pair something visual with [maintenance tasks]. think of IKEA, with the instructions. we try to pair those to workers"

SS: "reality first. we kept talking about reality and spacial. digital twins used to look like CAD models and be the source of truth, maybe my ERP system would be there, but the rest of the stuff wouldn't be there. so this is that scan down to an eight of an inch [shows 3d scan of plant]. but that doesn't give us information. we combine the visual and the spatial to give us the data all the time. we call it visual first"

PI: "if you separate the software from the data, they can live on after the job. digital twins -- we see the gaming side start to penetrate, the gas side start to penetrate"

SS: "we said data, data, data over again. don't stick yourself on one platform. we're working with Open Asset Digital Twin Working Group"

all of this is like 2% of the words spoken in this talk

there were slides

the "name an object in this photo" of talks

I used to think that people who did comp sci had to be smart. It seems like its filled with the stupidest people on earth

okay. buried underneath the babble was an actual thing, as far as i can tell. tl;dr [major agricultural company] started putting their 3D schematics, 3D photoscans, and regular-ass 2D photos into the same application so you can click around inside basically Blender, but for processing plants

which. cool. but they didn't actually specify a use case

Thats clever

they maybe hinted at possibly identifying flaws or defect in photos, either by human eyes or by machine learning, but as usual: "could" is a word meaning "doesn't"

[video ad for major agricultural company's "Digital Twin Platform"]

Bro what is this shit. It looks like one of those early 2000s 3d animated music videos. What decade is it???

These people are brainrotted

next talk. panel discussion actually. [Perry Olson] from [manufacturing organization], [Oscar Page] from [manufacturing company], [Floyd Atkinson] from [another manufacturing company], [Rodger Townsend] from [agricultural company]

title is "The Digital Inventory: How Additive Manufacturing Is Changing The Industry"

RT said something to the effect of looking to convert their physical parts... into digital twins (meaning schematics / 3D models / whatever)...? I can't tell if they're trying to dropship or something

FA: "we were able to get a part replaced in 24h because we had a digital twin [3d model][that we were able to print]"

okay, FA confirms they are trying to do parts-on-demand

("additive manufacturing" is the industry term for 3d printing)

it sounds like they're trying to do open-source, but for 3D models of parts

OP: "so the [shared platform] would have models, schematics, etc. from the OEMs accessible by companies with memberships"

okay maybe it's actually stock photo library subscriptions, but for 3D models of parts

RT: "just Google elon musk raptor 3 vs raptor 1 and compare the... how far it's come with additive manufacturing"

OP: "if a foreign company won't give you the schematic for a part or object you have, you can 3D scan this item and then you own it. it's your item now"

FA: "...because it was so impotent -- sorry, so infantile, we had to..."

spacex cant make rockets. theyre dogshit

next talk. [Firoj Umara], Asset Information Manager and Digital Twin Partner at [major agricultural company]

talk is "Asset Information Management"

FU: "we're trying to take away the fear of data, especially with this industry where data has grown magnanimously. we've all had to deal with feeding bad data into systems. and now we're dealing with our data being fed into AI systems. in this presentation i'll try to take away the fear of data"

meme onscreen of audience raising hands for "who wants clean data". audience lowers hands for "who wants to clean the data". audience gone for "who wants to pay for clean data"

points i guess for adding a third panel

these people dont know what the words they use mean

slide: "value realization from data is gaining traction now and hence 'data is the new asset'"

slide: "data is to an asset as blood is to the human body - keep it clean always"

slowly realizing this talk doesn't even have a concept that is being obscured by buzzwords, like the [other major agricultural company] thing. this is only metaphor

as in. "if your data is getting dirty, you need to identify sources of bad data like taps and turn them off"

and "if your data is bleeding out, you need to patch up the leaks"

all of those are metaphors but there is no underlying meaning behind them

wish i stuck with linguistics. this wouldve been fascinating as a study

FU: "good data plus AI is artificial intelligence. bad data plus AI is artificial stupidity"

unfortunately i cannot stick around for the three remaining talks:

• Knowledge Management and AI: Pivoting from Information to Insight
• Applying AI to Failures, Outages, and Turnarounds
• Technology for Improving Construction Throughput and Turnaround Efficiency

and of course, from [my regional manager] himself:

• Closing Keynote: Quantum Computing & AI: Utilizing Exponential Technologies

i have to leave because i have boycuddles :3