The CAD Insider

The New York Times Discovers Generative Design

Generative design, the overhyped and ultimately disappointing technology, one that was to have computers design perfect part, may have finally been discovered by the popular media – but only because it is being grouped under the public’s latest shiny object: AI.

In “35 Ways Real People are Using A.I. Right Now,” the 5^th real person is a NASA research engineer who has used a generative design program to make the lightweighted part for a satellite. Which generative design software we will never know, because the  venerable publication succumbing to the fever over the hottest tech story in years (ChatGPT and AI) attributes the design to AI.  Not once is “generative design” or “topology optimization” used in their article.

It’s not like the Times doesn’t cover technology. They have a Technology section. The paper’s Wirecutter reviews of tech products are great plain-English information that can boil an ocean of products into one or two recommendations.  To be fair, they did cover generative design 2 years ago -- though it was only for its novel use in jewelry design.

Should we even expect the popular media, even the best, to go deep into technology? The Wall Street Journal, does a commendable job on the business side of tech, even brushing on the science. 

I shouldn’t complain. The lack of depth in real technology coverage is an opportunity -- and job security -- for tech journalists.

It’s Easy to Use if You Know How to Use It—The UI Change Dilemma

An overhaul of the user interface is necessary after some time but poses a dilemma for every software company. Just as a home could use a fresh coat of paint to avoid a tired, dated look, so could software apps. Every software company knows but puts off big UI changes because that is a)expensive and b)could cause a riot.

AutoCAD users practically rioted when Autodesk changed their command line driven UI and took away their shortcut-filled tablets of a DOS-based system to a mouse-driven interface that matched the Windows look and feel. SolidWorks users grumbled a little. Heard was “don’t you guys have anything better to do? Like fix the crashes?” But having grown up on Windows, they didn't riot.  The SolidWorks UI team was able to get away with modernizing the UI with better icons, less clutter, etc. Let's face it. SolidWorks had grown to do so much that rows and columns of icons crowded the screen and turned their working area into a peephole. None of the functionality was lost with the UI changes but it still took a little getting used to. Users were annoyed to have to hunt for their favorite tools.

PTC faced huge changes going from minicomputers like the DEC VAX to the PC and from Pro/ENGINEER to Creo. But it all seemed to be contained. I did not hear of Pro/E users leaving the flock in droves.

A UI refresh is a huge undertaking. But when announcing it, one has to be careful. You may want to blare the trumpets and reveal the results of the massive undertaking, so proud are you, but then you have all the users who have gotten  by now, are quite used to your software the way it was.

Don’t even think of telling the users that the new UI will make the software easier to use. For the users who knew how to use it, anything new will be harder to use.

Volume Graphics, makers of VG Studio, a CT scan processing application, has chosen to release their UI update (read about it here on engineering.com). Although it has a radically different marching menu, the change comes with version 2023.1,  a point release. It’s a genius approach. It will look as if the company is saying the UI change is no big deal, nothing to be scared off, not like a big annual release -- while still having a thoroughly modern look in their current release.

For whom a fresh coat of paint is important, i.e., new homeowners, the clean new look means everything. This house may be old (that’s good, it’s built to last) but look how well it’s been kept up.

And this Month the PLM Leader is… Siemens

Siemens is busy congratulating themselves on Twitter for being the leader in PLM, citing a Data Quadrant by Software Reviews that placed them far and away better than their competition in terms of “product features and satisfaction” and “vendor experience and capabilities.” Far ahead of Dassault Systèmes’ ENOVIA,  Autodesk (though Fusion 360 is not a PLM product), Arena (SaaS PLM acquired by PTC) and even further ahead from PTC's Windchill.

Only late last year, PTC CEO Jim Heppelmann joyfully discovered his company was the the PLM market leader after combining the number of users of Arena and Windchill.

Not much is known about SoftwareReviews. Are they yet another algorithmically judgmental all-purpose social media scraper of comments from which product quality is somehow derived? Are they legit, basing their result on the collective considered reviews of bona fide PLM users, or industry experts (like Gartner with their Magic Quadrant) or – perish the thought—journalists?

Another review site, G2 reviews, has been used by so called winners as proof of their superiority. G2 also has their own quadrant for PLM and also has a Siemens Teamcenter as a leader in "market presence" but practically even with Arena PLM in "satisfaction."

ChatGPT Hijacks a CAD Conference

(Image generated with DALL-E 2 using the title of this blog entry as the prompt. Quote is mine)

The word of the conference was to have been “imagine” but all the chat was about ChatBot. The AI chat sensation derailed one presentation after another. Presenters gave up their subject to talk about AI and ChatGPT – not because they were being pestered by a rude press corp, but willingly. anticipating the questions that never came. They had presumed that it was the only thing on the media’s mind. It’s easy to see why. Every day leading up to the conference, they had been showered with article after article in the popular and business media about ChatGPT.

After one presenter volunteered the subject of ChatGPT, a  journalist dared to ask how close we were to the dreaded dreaded Singularity, referring to the point at which AI can starts improving itself, quickly becoming smarter than its creators. We, the people, no longer of any use to this super intelligent sentient being, are eliminated.

That’s just science fiction, right?

ChatGPT is the two month-old baby of an San Francisco AI startup no one paid much attention to until Microsoft gave it a $10 billion dollar present. Suddenly, it was the cutest, smartest baby in the world.

But ChatGPT is more like a con artist than a baby. You want to believe what it says because it looks so believable. Its answers are well formed, grammatically correct. Once it takes human form, it will most certainly have an English accent and wear a suit.

Like a con artist, ChatGPT lies. It’s answers to questions were incorrect when it was introduced and continue to be incorrect an alarming number of times. If you are using ChatGPT for research or for engineering information, you had better check every answer.

You would think journalists and engineers, other who pride themselves in truth and accuracy, would consider themselves safe from being made obsolete by such a charlatan.

The technologists move to allay the fears of obsolescence they presume are in the room . Fear is only the first stage in the acceptance of any revolutionary technology. To prove his point, they give an example.

“I asked ChatGPT to generate code for [sophisticated geometry] and it generated perfectly structured code.”

“Did it work?” His colleague asks.

“No. But I fixed it.”

He was trying to show how useful ChatGPT could be, how a developer could have used ChatGPT to get 90% of the way there.

I tried to get ChatGPT to finish this article but I was not nearly as successful.

I guess my job (journalist and engineer) is safe. Mostly. For now.

Google Stumbles in the Race for the Perfect Answering Machine

The anticipation for Google’s Feb 8 event was high. Alphabet, Google’s parent company was to deliver its answer to ChatGPT, the precocious chatbot that has AI-addled world in a tizzy with its ability to answer college-level essay questions, program code, name your dog... and God knows what else.

But Google did not deliver. A beta version of its chatbot was still a month away and still not ready for public testing. We do have a name, though. Bard. After Shakespeare, I presume. But wait, don’t go, they said. Look at all the marvelous things we've done with AI.

A series of innovations Google had previously introduced was listed in case you had missed them. Google does tends to slip things in and to give you joy on discovering them, without fanfare. Like its calculator, which I discovered accidentally can do all sorts of math and plotting. It was never announced, never productized and still doesn’t have a name, it’s interface still only the search box. To enjoy it you have to know about it. It’s like the secret menu at In ‘n Out.

We gave you Google Assistant years ago, says the Google presenter. It’s practically a ChatGPT plus voice activation. And don’t you love the visual search, the presenter asks. We have made the camera the keyboard.

I did not know about it. They never told me.

And Google Maps has an immersive view, they say.

I try to find it and cannot. It must be available for Android phones. I have an iPhone.

“You want something new?,” they say, perhaps remote sensing my interest waning through my Google smart speaker. “We have an app for arts and culture.”

I’m not sure anyone asked for an app for arts and culture from Google. It sounds like a pet project by Google engineers wishing they knew a little bit more about paintings and opera and assuming everyone else did, too. .

A presenter reaches for her smart phone (no doubt a Google Pixel) but can’t find it. I died for her. But it was hardly the worst fail in live product introductions. Remember the blue screen of death when Bill Gates was introducing Windows 98? Or when they shattered the “bulletproof” glass of Tesla’s new Cybertruck by throwing steel balls at it? “But it didn’t go through,” says Elon Musk.

Why do tech companies insist on live events, I wondered and laughed. Elon Musk doesn’t need my pity.

By the time the Paris event was livestreaming, the damage had already been done. Alphabet’s CEO Sundar Pichai had announced the company’s chatbot, Bard, with a sampling of questions that Bard could answer. It should have been a master stroke, the search giant able to field and correctly answer questions spanning human knowledge, the robot equivalent of Peter Jennings (Jeopardy! Champion) with the vast reach of Google, or the head reference librarian at the Library of Congress. Except one of the answers was wrong. So stated Reuters, the news agency that prides itself on getting it right, which had done the fact check that Google, in its haste to pull even with ChatGPT, had not. The James Webb Telescope didn't discover exoplanets. The European Southern Observatory's telescope did. It was a mistake that could easily have been avoided if Pichai had used Google instead of Bard.

Bad news travels fast these days. Investors were quick to punish Google for the blunder, sending Alphabet shares down 9% -- a loss of $100 billion in market value—before they evened out.

More important and more permanent may be how this reflects on Google. The uncontested search leader for 25 years, has Google gone from novelty to yesterday’s news? Has Google become the new Microsoft, a company whose products you use all the time, even depend on, but have lost the luster of the shiny new objects?

Google still has the best search, bar none. It has taken great care in sorting the Internet, referring us to safe sites. We can be sure Google will not send us to a site bent on swindling us, using copied articles or one that has spoofed its way to the top of the search results.

Google had resisted introducing its chatbot until that fateful Wednesday because it was not ready for public use. Google is now an established company with a reputation to preserve. Bard was unnamed and being tested. Wisely, it was held back to avoid the risk of an embarrassing failures of chatbots past. In 2016, Microsoft’s chatbot Tay made it sixteen hours before spewing a series of racist and racist and misogynistic tweets and Microsoft had to shut it down. Meta (formerly Facebook) yanked its chatbot Galactica after it “spat our biased and incorrect nonsense,” according to TechnologyReview.com.

Google is trying to be more “responsible,” as it stated 4 times during the livestream.

No doubt Googles leadership panicked upon hearing that ChatGPT landed a $10 billion investment from its arch enemy, Microsoft. It was easy to see how Microsoft could, with ChatGPT’s technology, make Bing, the most expensive search engine nobody uses, to one that leapfrogs Google.

Pichai, in a panic, issued a code red to his managers and called Sergei Brin and Larry Page, Google’s founders, back into the building for an emergency meeting and followed quickly (too quickly) with his fateful product announcement – overpromising the accuracy of a product still at least months away from prime time. It would have been wise to wait, issue a more reserved statement, as “we are working on it” sort of thing, and continue with having Bard learn and testing it, making sure it can provide the right answer, say 99% of the time. No company besides Alphabet has the resources to do that or such a head start in search and AI.

A little patience would not kill them.

Should ChatGPT be Renamed CliffGPT?

Should OpenAI adopt the face of Cliff Clavin, the mailman from Cheers, for its ChatGPT? Cliff always had an explanation for everything, but it was not always right. Here is a picture from YouTube video of Cliff on Jeopardy! on an episode of Cheers.

ChatGPT exists in a question and answer format and for it to be more human, could use a face. No better face than of Cliff Clavin, the lovable know-it-all, a regular at Boston’s fictitious bar, Cheers? Then OpenAI (makers of ChatGPT) could call it CliffGPT. Like Cliff, ChatGPT is great entertainment, great for being initially impressive in some some subjects but ultimately proves unreliable and untrustworthy in mission critical subjects. Like engineering. I tried. Read about it here: ChatGPT Has All the Answers – But Not Always the Right Ones.

You should not design a plane or a dam around CliffGPT answers.  It’s knowledge and reasoning may have let it pass medical exams but I doubt if could pass the EIT exam and pretty sure it would bomb the PE exam.

But stay tuned. Microsoft just pumped $10 billion into OpenAI. ChatGPT is only going to get smarter, wiser...

Water Shortage is One Thing, but Lights Out?

(Image from Wikipedia)

It may be hard to see thousand-year trends in a daily newspaper, but that doesn’t stop me from trying.

The WSJ is reporting on the US Bureau of Reclamation cutting back the use of the water from the Colorado River. That’s part of the two trends. The first is too many people (overpopulation) and too much heat (climate change).

The Colorado River starts near Denver, where it picks up its name before slices off the corner off Utah and snaking through the Grand Canyon, separating Arizona from Nevada and California before it dumps into Mexico’s Gulf of California. Except there hasn’t been any dumping for many years now. The US states have damned the river and siphoned off so much of the river water for irrigation and consumption that none of it is left to flow into Mexico.

The US West has been baked with heat waves. This year, we got some rain but a lot of the water evaporated in the heat. Lake Mead, created by the Hoover Dam on the Colorado River, is at the lowest level since the dam was built.

With every state helping itself of Colorado River water, the federal government had to step in. And so, we have been put on rations. Though not all of us, though. California, on the edge of the massive Colorado River basin but its biggest user, is left out of the Bureau of Reclamation’s recent limits.

If you’re thinking “So what? I can flush my toilet less and get bottled water,” think of doing that in the dark. If water levels go much lower in Lake Mead or Lake Powell, the hydroelectric power generation stops completely. There go the Western power grids.

What Exactly is a Digital Transformation?

“Digital transformation” may be the latest industry buzzword. The Big Four CAD vendors (Autodesk, Dassault Systèmes, PTC, Siemens) and other software vendors insist every manufacturer needs to undergo a digital transformation, and in doing so, will gain a competitive advantage since it’s still the early days of digital transformation adoption. But ignore the need for a digital transformation for too long and your company will be hopelessly left behind.

But what exactly is a digital transformation? Perhaps your company is already undergoing a digital transformation at its own speed and in its own way, and not even know it is transforming according to what has been prescribed by the technology vendors. Maybe a digital transformation is just a “hurry up” from technology vendors who are growing impatient for its adoption?

It would not be the first time software companies have coined a term for products they sell. Computer aided design, computer aided manufacturing and computer aided engineering (now more commonly called simulation) were offered by software companies as solutions for problems manufacturing companies did not know they had. CAD, CAM and CAE are now commonly accepted as genuine needs. CAD, CAE and CAM were followed by PLM -- at least by by Dassault Systèmes, PTC and Siemens. Autodesk held out, declaring as late as the Carl Bass era, that PLM was not really a thing.

“The only people who need PLM are the ones selling PLM,” Bass quotes in Kenneth Wong’s record along with “PLM is a solution in search of a problem.”

Digital transformation, however, is in more need of definition than CAD, CAM, CAE and PLM – if only because there is no single software application to supply it. It still floats into a conversation as vague concept, but one that demands that you, if you are at all tech savvy, should know about it.

Autodesk, a vendor serious about digital transformation, convened a Who’s Who of industry consultants to make the concept clearer at their Accelerate 2022 conference. Onstage and on the spot, the consultants valiantly attempted to agree on a definition of a digital transformation, on its need and on its implementation.

It would have been nice if they would have established milestones so companies could judge how complete was their own digital transformation, but this was only an hour-long discussion. This was not a formal standards committee. And, let’s face it: a  Autodesk conference was hardly the place to quibble about the necessity for a digital transformation when your host has requested that you discuss its merits.

The Shape of Things to Come

Stephen Hooper, Vice President & General Manager, Design & Manufacturing Cloud Solutions at Autodesk, is about to reveal what is coming up in generative design. He is at a Q&A session at Accelerate 2022 conference. He is not making an official announcement as the technology is still in R&D. If it were up to the PR watchdogs in the room, he would go no further. But Hooper’s enthusiasm is not to be bridled.

“We’re going to use ML [machine language] to look for similar shapes,” he says about the next generation of Autodesk’s generative design, which will, no doubt, be manifest itself in Fusion 360, the company’s do-all design and manufacturing platform.

We will be able to create a 3D map of your shape and match it with another existing shape, he states. Engineers reinvent the wheel all the time, he implies.

We asked if the generative design of this future could not only look for similar parts, but look for similar elemental or primitive shapes, such as I-beams or round tubes, the shapes engineers rely on.

Yes, says Hooper. It certainly will.

Hooper, whose youthful figure belies decades of CAD experience, tells of his initial introduction to CAD with Unigraphics. His conversion to Autodesk came when he realized that, despite Unigraphics superiority in CAD, its expense was a hindrance.

“For what we paid for a few UG seats, we could outfit a hundred engineers with Fusion 360 today,” says Hooper.

He invokes a Moneyball reference, the movie with Billy Beane (played by Brad Pitt) creates a pennant winning major league franchise (the Oakland A’s) by recruiting underrated (and therefore underpaid) players and creating an overall better team when tradition was to win with one or two (very expensive) superstars. While this could be considered an admission of PC-based AutoCAD’s inferiority to the workstation-based CAD, the point Hooper is making is consistent with Autodesk’s humble approach to democratizing design, simulation and manufacturing software.

Only 3% of Us Want to Work from Home? Can this be Right?

Only 3% of employees want to work from home 100% of the time, says Kim Dabbs, Global VP, ESG + Social Innovation at Steelcase, citing her company’s survey at Autodesk’s Accelerate 2022 conference currently taking place in New Orleans. Can this be right? Most employees we hear from are fine where they are, mostly working from home except for going to the office “as needed.” That’s code for “as little as possible.”

Consider the source. Steelcase exists to make office furniture and fixtures and a permanent and complete change to a work at home status for workers in their markets is an existential problem. Steelcase is asking the question and presenting the result in a manner that is to their advantage. The same question to a US market increases the figure of employees wanting to work from home 100% of the time by 40% (5%). Asking the same demographic whether they want to work at home most of the time may yield a result much higher, more in keeping with the common perception – and less advantageous to a manufacturer of office furniture.