AI Will Make Electronic Design More Valuable to Our Readers

There’s a storm coming

Of mistruth and deception

We’ll weather it out

In his recent article, "AI in the Tech Industry is Here—and It’s Reshaping the Job Landscape," my colleague Cabe Atwell portends the changes needed for Electronic Design’s readers in the tech business to survive the coming of AI. He states how adaptation and learning to work with AI tools are becoming the key to staying relevant in the tech industry. The piece goes on about layoffs already happening and how everyone’s getting excited about the AI purveyors’ promises of increased efficiency through jobs elimination. According to Cabe’s article:

“Goldman Sachs has said that AI could replace the equivalent of 300 million jobs, citing shifts in workflows, which the banking firm ultimately states could drive up global GDP by 7%.”

In response to Atwell’s negative-Nelly AI prognostication directed at the tech business, one of our readers said in the comments section of his article:

“Not just the Tech business, journalism is also surely going to change. The move from Paper to digital killed off a lot of jobs, now when you can simply type "Compile me an article on the latest developments in Microprocessors" why would you need a journal?” — Martin OHara.

Of course, being built the way I am after several decades as an EE, I typed Martin’s prompt into Google and got a rather watered-down overview of the latest microprocessor developments. Interestingly, some of the subject areas were generically described, while others mentioned ONE representative vendor in each area. As if only one company was innovating in each topical area. Leading me to ask myself, “Why THAT company?”

Those of you who have been around for a while, like I have, probably remember the annual Microprocessor Directory that Rob Cravotta used to produce (he also did a DSP directory) over at another “journal” that shall not be named. Rob’s directory was a comprehensive gem and listed pretty much everything that was available that year, if I remember it correctly, and was a labor of love in terms of the details provided in organized columns.

For instance, the reader could hunt down all of the devices that had 16-bit DACs, and then 16 bit ADCs, and eventually arrived at a narrowed, UNBIASED selection from a comprehensive printed database of components — one that would survive a burst of EMP.

Billionaires Buying Power

So, here’s the thing: Agreements to acquire entire nuclear reactors at places like Three Mile Island are being strategically inked to supply the gigawatt levels of power needed by AI data centers. AI will be the new knowledge worker, the librarian, the search engine that consumes 10X the power of regular search to provide employed humans(?) with the information needed to fire another sector of humans that were putting food on their families’ tables. Those newly unemployed won’t benefit from the leisure provided by AI.

“I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes” — Joanna Maciejewska, author and videogame enthusiast

Nor will they benefit from a Universal Basic Income because of society’s productivity gains. That all gets pocketed by a dozen or so already-rich dudes, most of whom seem to like to pretend they’re Lex Luthor in their off hours.

These AI search tools are based on training data, pilfered copyright materials, copyrighted artworks, and from absconding with expertise found in subject-matter-expert collectives like tech forums, LinkedIn, and Reddit. I recall a point in time, recently, where LinkedIn was posting arbitrary technical questions, which provided a gyre for subject-matter experts to gather around and where they could show off their knowledge to lurking talent scouts and headhunters. Little did they realize they were training an AI to be sold off as training data to eventually take their jobs. Smart, but dumb, humans.

Heck, all of the stuff I’ve been posting here on Electronic Design gets scraped by AI, training it with fresh knowledge, including the yet-unheard-of breaking news in my April Fools’ article on fueling hydrogen cars at Tesla Supercharger stations. OMG, I’m not only training AI to do my editor job, but I’ve also trained it to sucker the gullible, and to incrementally move toward deceiving humans with a convincing weave of facts into its deadly web.

“Will you walk into my parlor?” said the spider to the fly;
“‘Tis the prettiest little parlor that ever you may spy.
The way into my parlor is up a winding stair, And I have many curious things to show when you are there.”
“Oh no, no,” said the little fly; “to ask me is in vain,
For who goes up your winding stair can ne’er come down again.” — Mary Howitt

So, back to Martin’s point: Will AI replace me and the magazine we’ll have enjoyed for 75 years in 2027? I don’t think so...

Long Division

Since its earliest days, the internet has always stooped to scraping the bottom of the barrel of humanity, the lowest common denominator. It provides communities, for each village’s one idiot, to emigrate to and in which to reinforce each other’s ignorance, conspiracies, deception, and fraud, as well as build a convincing, Dunning-Kruger-based confidence in their expertise (climate change is THE poster child). The internet still had the best of humanity among all of the noise, though, and was a wonderful research tool back a couple of decades ago.

Google search has, sadly, devolved to where it is next to worthless today. Type in a complex word, and you get a full page of a movie, or airheaded video game, instead of the definition of the word from Webster’s like we got a mere few years ago. AI has trained on this current manifestation of the internet, a pathetic shell of what the web was a couple of decades ago.

We can learn from what happened with classical search in terms of where AI-based search will go. Back in the old Netscape browser days, content wasn’t fully developed, but it progressed to where search produced credible results. The most relevant results were delivered, with ads occupying the first two or three returns, well-tagged as an ad. This evolved into the commercially ranked mess we have now, where Google usually delivers complete trash, full of irrelevant, commercial content, which is very frustrating.

These days, we’re being played psychologically, where the top-ranked, biased, paid-for search that’s now mostly irrelevant is preceded by an “AI Summary.” The AI Summary seems to deliver more relevant results to the search query. It’s presented in the digestible form of a few paragraphs or bullet points, along with links to the material upon which the summary is based. You’re being herded to use the AI search.

Preying on Laziness

To Martin’s point, if you can get a Google AI summary of what Andy wrote last week, why would you read Andy’s Electronic Design content? The primary reason, of course, is that the AI summary isn’t detailed, cannot reason, has zero hands-on experience, and has little subject matter expertise. And if you don’t have the sufficient depth of subject-matter expertise to assess the quality of the answer, you go to the reference material being cited.

These summaries are getting better, and, more times than not, searchers aren’t happy with classical search relevance. Instead, they’re actively clicking on “AI Mode” for more relevant results as a few paragraphs of summary, along with that set of reference links where more details can be followed up.

Publishers lived and breathed on search — material has to be found to be read. This was traditionally based on relevance; then the art of SEO (search engine optimization) played rulesets that got higher rankings in search than competing materials.

Like Mad Magazine’s Spy vs. Spy, a search war escalation ensued where Google would try to see past the SEO games by changing its search algos. At that point, SEO specialists and consultants would come up with new tricks to raise those declining search rankings, getting fewer visits to the publisher’s content.

With AI search, the need for a visit to the reference materials is, to Martin’s point, almost pointless. Rather than reading Andy’s 1,200-word article, you get what you want from the AI search summary in two or three paragraphs.

So, “why would you need a journal” like Electronic Design? A valid and seemingly scary question, Martin. At the risk of sounding smug, I’m really not that worried, because human behaviors have remained the same for millennia — only the tools and technologies have changed, cycling through those behaviors.

History, and Search, Repeats Itself

AI search is a tool and a technology. We’re at the same point now with AI search as we were in 1998 with Google’s new search engine. It was a useful tool. We learned to use it, to trust it. We made it an essential part of our research and design process as engineers. No more trips to the university library. You could research patents, papers, and companies from your desktop computer and at speeds well beyond the 2400-baud dial-up I had at the start of that decade.

We had the rich format of a web page, not the eerie green glow of ASCII characters on a VT-100 monitor. We tolerated the mild commercialization of the results on the search page, for it was a clearly labeled “ad” and it only occupied a couple of positions at the top of the page — easy to ignore.

Once trusted, relied upon, and essential, the commercialization of the search results then crept in, where search results ranking could now be bought or manipulated to produce trash. It was forcing us to find relevance two or three pages into the search results. In the last couple of years, search has been almost worthless with Google, and those of us doing research had other search engines we’d use, beyond merely finding the phone number of the nearest pizza joint.

Enter AI search. Now we’re back to the 1998 search behavior again. AI search delivers what we’re searching for AND it explains the results to the point where we don’t have to dive further. Those of us with a modicum of competence can smell a large language model (LLM) rat that was delivered to us as an AI summary, so we know to look in detail at the cited references.

I made that point in my last blog, where CFOs appear to be under the mistaken impression that they can hire low-wage, unskilled workers as “prompt engineers” to replace experienced knowledge workers — unskilled people who can’t determine the validity of AI summaries.

These people are all over communities and forums nowadays, armed with ridiculous ChatGPT-suggested schematics, drawings, and instructions. They defend ChatGPT results when human experts expose the ridiculousness of the solution and offer advice or insight (inadvertently training and refining AI in the process, like they did with my April Fools article; Perplexity repaired its gullibility within an hour or so).

AI is getting better by the day. It seems to produce more credible summaries, reducing the need for visits to tech journal sites. Is Martin’s crystal ball accurate? I’d say, in the short term, for a couple of years, it appears grim. But...

AI is now in the phase of developing trust, like Google did in the late 1990s, and it’s intertwining itself into our processes and organizations that rely on the AI as an essential functional element. Amazon, Google, Microsoft, Perplexity, and others appear to be benevolent. They provide us with trusted search and summaries, appear to be experts, and are spending $billions on building data centers so that we can get such outstanding search results and functions, they promise to replace workers — all for $20/month. Nice?

No. Don’t forget that Facebook and others have used psychological games to create addictive behaviors and manipulated what you saw back in rev 1.0 of this mess we’re sinking into. I think that once the trust is established, once they entangle themselves into the corporate world (that includes the dangerous proliferation of AI in U.S. government agencies), once they starve out those journal references they now cite and replace them with material written by another AI, those AI search and knowledge engines can be manipulated, much in the same way as Google, on behalf of the highest bidder.

We’ll get to the point where AI search gets monetized ranking, similar to when I typed in Martin’s prompt and got ONE semiconductor company as the example. Can I prove that it was a paid AI result? No. Could it be? It's the way to make boatloads of moolah, so, even if it’s not, it WILL BE.

And this is where we come full circle. Just like traditional search has now frustrated us to where we allow ourselves to get herded into AI search, AI’s relevant results will become corrupted, filled with paid prioritization and, worse, with seeds of misinformation that suit special interests in influencing the masses.

AI Has No Firewall

It’s at this point where we’ll seek out the subject-matter experts, the ones whose newsletters we’ve kept coming into our inboxes. Because, you see, there’s a firewall between editors like myself and my colleagues, standing between us and the sponsored (paid ads) content side of the business. We write about companies and products because we like what we see as seasoned editors, subject-matter experts, and as engineers, and we curate that content to be interesting for our readers.

Much like Rob’s Microprocessor Directory, we give you the info, the analysis, and ask questions on your behalf. YOU get to decide what you see and use, not an AI that’s been bought and is filtering out the highest bidder’s competitors.

Bit of a storm coming up quickly for publishers. We’ll see many publications sadly get sunk. But I think as long as we keep in touch with our human readers via our newsletters, and we get emails and comments from them to promote human to human discourse, and as our execs batten down the hatches and ride this one out, we at Electronic Design should weather it just fine.

In my vu, anyway,

-AndyT

Andy's Nonlinearities blog arrives the first and third Monday Tuesday of every month. To make sure you don't miss the latest edition, new articles, or breaking news coverage, please subscribe to our Electronic Design Today newsletter. Please also subscribe to Andy’s Automotive Electronics bi-weekly newsletter.