Google said the Panasonic Word Processors did not self test mode. Look what I found...

[-]

rin3y@reddit

Might I recommend this Chrome plugin that hides Google's idiotic AI "suggestions"? https://chromewebstore.google.com/detail/hide-gemini-and-google-ai/

[-]

TimChuma@reddit

Don't trust anything google AI results say it is 100% garbage

[-]

tes_kitty@reddit

The problem is, it's not, it will be right enough times that people start to think it's right every time. And that's the problem.

[-]

istarian@reddit

Which is called 'being stupid' because it can work exactly the same way with interactions between humans..

[-]

istarian@reddit

You want to expand on that?

Because I do not agree with you on this one.

Lots of people place far too much trust on other people on the basis of things unrelated to true subject matter expertise.

AI will happily and confidently give you incorrect information because it doesn't actually know whether it is right, this is true. But in reality people also do the same thing, confidence != correctness.

[-]

tes_kitty@reddit

The difference is that people have been conditioned over a long time to trust the output of a computer since you could trust that if you shove numbers into an equation (in Excel for example) the output generated by the computer will be correct. Same for looking up existing data in a database.

AI changed that, suddenly the output, while still sounding correct can be completely made up.

(Yes, programmers know better, but once their software has been debugged and reaches the public the above is mostly true)

[-]

TimChuma@reddit

I have to fight people on the phone who believe it "because its on google" well it is not correct. Also the price is what is on our database as thats what the shop charges

[-]

nobody2008@reddit

Nice! Also, Google AI = Some parsed answer from some random person on a forum somewhere.

[-]

Dense_Occasion9971@reddit (OP)

I wish there was a mechanism to send corrected information to Google.

[-]

LifeguardGlobal952@reddit

This will most likely form future results, do t forget to add the answer.

[-]

banksy_h8r@reddit

Why? They aren't paying you to make their product better. A better contribution would be a webpage or wiki article with authoritative information and let all the search engines pick it up if they care.

[-]

takeyouraxeandhack@reddit

AIs will always hallucinate. The way they work makes it impossible for them not to hallucinate.

[-]

DL757@reddit

Simple, just ignore the AI

[-]

TygerTung@reddit

That Google search AI is so unreliable I don't trust it at all.

[-]

tpimh@reddit

Often unrelated to the search topic

[-]

classicvincent@reddit

It blows my mind that people trust the answers that they get from google AI search.

[-]

istarian@reddit

I mean they also trust the answers they get from friends and family...

[-]

Hjalfi@reddit

I found the service manual --- this is powered by a 6303! (aka an extended 6800 made by Hitachi). Which is probably only interesting if you're into weird CPUs, but luckily I am...

https://cowlark.com/2020-10-19-6303-arithmetic/index.html

[-]

Kawashiro_N@reddit

So like the TRS-80 MC-10.

https://en.wikipedia.org/wiki/TRS-80_MC-10

[-]

istarian@reddit

Not exactly, but I'm sure they share many similarities and you might be able to build from the same source code with a few modifications.

The TRS-80 Color Computer(s) use the Motorola MC6809 and the MC-10 uses a Motorola 6803. Hitachi made the 6309 which is compatible with Motorola's 6809.

[-]

midnight-salmon@reddit

AI "answers" aren't answers in any real sense, they're generated text matching heuristics to give an impression of an answer and nothing more.

[-]

istarian@reddit

At least some of them are answers in exactly the same sense that something you learned in school or from a book is an answer.

The problem is that AI is akin to a small child spouting nonsense from a questionable source...

[-]

JustHereForMiatas@reddit

If it's undocumented, or poorly documented, google won't know about it. From their AI's "perspective" of having to give a confident answer to every question it's asked, it'll just confidently tell you the wrong thing.

In my experience, these things usually have some kind of hidden test mode somewhere. For example, I have a Sharp Intelliwriter PA-1000H. Looked online for a documented test mode, and there wasn't any to be found. I managed to find it anyway through trial and error.

And just on the very, very off chance you need it: for the PA1000H, you hold down Code + Z + C while powering on for the diagnostic mode, and (this one was documented) Code + 1 + 2 while powering on for the demonstration mode.

[-]

LaundryMan2008@reddit

You can also stumble upon it by accident, did so with a leap pad and my phone, the leap pad I was bored pressing random buttons but my phone I was trying to turn it off fully but didn’t know how to and I pressed some combination of the power and volume buttons which entered an Apple factory test mode, funnily enough it had the power off fully setting so I did that anyways before leaving to go to lessons

[-]

JustHereForMiatas@reddit

Oh yeah that's sometimes the way.

With the PA1000H the first time I encountered the screen was when I was messing around with its proprietary, unused printer port. I shorted two pins together by mistake and it went into diagnostic mode.

It turns out that that was a completely random fluke that I couldn't recreate... but now I knew that there was a diagnostic mode, and had to find it. Already knowing about the store demo mode and that it was accessed by holding buttons on power up, I went hunting and eventually found it.

[-]

AmoreLucky@reddit

This is why ChatGPT isn't to be trusted, it's literally just an app that strings random words together regardless of how true they'd be or how much sense it makes. Also, it's so cool seeing early word processors like this, I'm betting some college students made some good use of these machines back then whenever they had access to them. \^_\^

[-]

2raysdiver@reddit

On more than one occasion, when I have looked at the link(s) to the right of what the Google AI says, the referenced website says exactly the opposite.

[-]

thunderbird32@reddit

Well, ChatGPT is pretty bad at more obscure stuff like this, and when you say Google I assume you mean Gemini? Same deal there, if so.

[-]

Dense_Occasion9971@reddit (OP)

Yes, Gemini. Google search bar. ChatGBT. I've used them all.

[-]

Romymopen@reddit

See what happens when you contradict the hive mind?

AI = bad

Well, I used Google Search too.

DOWNVOTE ANYWAY

[-]

thunderbird32@reddit

Ignoring any ethical concerns (with regards to training data or resource usage) for the moment, AI is a tool. Like any tool there are things that it does well and things it does poorly. This is one of those use-cases for which it does poorly.

[-]

Romymopen@reddit

You're preaching to the choir. I see no qualms with using every written word available to train our future overlords.

If you don't want your thoughts absorbed by people and computers, keep them to yourself.

We live in a society that does everything it can to create these giant companies. And they can't wait to empty their wallets into Jeff bezo's hands and into Steve Jobs' casket.

And then out of the other side of their mouths cry foul that those companies have tons of money.

And, not only support but, LOVE copyright, trademark, and patents. Get rid of IP laws and all those companies collapse.

No offense, but idiots, the lot of them.

I'm gonna use every tool I can get my hands on. LLMs, hammers, Facebook, coping saws, and whatever else helps make my life easier.

[-]

starcube@reddit

They all hallucinate and get a lot of shit wrong. It takes longer to use them and then fact-check their output than just do the work yourself. They're not really AI, they don't understand what you're asking them or what they're telling you.

[-]

TheThiefMaster@reddit

Gemini and Copilot both cite sources so as long as you actually follow those links it's not too bad to fact check them.

[-]

starcube@reddit

Why should I, though? The technology is inherently untrustworthy.

[-]

FarOutJunk@reddit

Well stop. You’re not accomplishing anything.

[-]

GaiusJocundus@reddit

Stop supporting llm's

AI is cool but LLM technology as implemented on the market today are destructive, plagiarism machines operated by owning class idiots who will crush you under their heel for a dollar and tell you to be grateful to have touched their shining sole.

[-]

Acceptable_Policy705@reddit

Horribly inefficient power-wise, too

[-]

QuantumCakeIsALie@reddit

They're all crap. Ask them the time is going to be in 5 hours.

[-]

marhaus1@reddit

Please explain how you got to that, someone will be googling about this in a few years and curse at his screen that you didn't give the details otherwise 😅

[-]

Dense_Occasion9971@reddit (OP)

Read my response to Eaglebtc. I give the secret formula. I forgot to include it in my original post and Reddit does not allow you to edit after you post.
.

[-]

eaglebtc@reddit

So ... are you gonna actually share the key combo and document it for the rest of us?

[-]

Dense_Occasion9971@reddit (OP)

Of course! Begin with the machine off. Then at the same time press the POWER button, the letter T and the key with Arrow Up with the little line across the top of the arrow. It's to the LEFT of the space bar. The diagnostic menu will appear on the screen. Press the number of the test you to perform and hit the return key.

[-]

eaglebtc@reddit

Dear Internet,

Today, OP delivered.

[-]

Dense_Occasion9971@reddit (OP)

Yes... I forgot to include the "How To" and Reddit wouldn't let me edit my original post.🙄

[-]

ATK80k@reddit

I'm not an expert but I am reasonably sure that all stuff like that has a test mode

[-]

Dense_Occasion9971@reddit (OP)

Interestingly, the manual makes no mention of this ability. It's buried in the Service manual.

[-]

leadedsolder@reddit

So how do you get to it?

[-]

joshsmog@reddit

Lol commenting just for someone in the future to find this thread thinking theres a solution and op decided to bait them. Ya got played son

[-]

Dense_Occasion9971@reddit (OP)

I just posted the combination of keys you need to press. See my response to "Eaglebtc". Reddit won't let me edit my original postm

[-]

ATK80k@reddit

Yes. When in doubt, or for the best solutions, hunt down the service manual for stuff!

[-]

busytransitgworl@reddit

And how did you get into that self-diagnosis mode?

[-]

Dense_Occasion9971@reddit (OP)

Just posted the key combination. Scroll down to my response to Eaglebtc.

[-]

therezin@reddit

Remember, LLMs don't answer your question. They implicitly answer the question "what would an answer to this question look like?". It will never actually check for correctness.

[-]

ziplock9000@reddit

Oh wow, you found AI lies.

Welcome to 2021

[-]

cazzipropri@reddit

Gemini doesn't know shit.

[-]

olifiers@reddit

The future we were promised!

[-]

t_Lancer@reddit

an another "stop asking chat bots unless you can fact check their answers!"

[-]

cthart@reddit

TYL Google doesn't know everything.

[-]

deelowe@reddit

I find a lot of times this is due to how the search query is written. Google has gone through some major algorithmic changes and it seems that talking to it like a human now works better than the more traditional ways of writing queries.

Here's the query I wrote:

Does the Panasonic W-1500 Word Processor have a self test mode?

And here is the response I got back:

While there is no information confirming a self-test mode for the Panasonic W-1500 word processor, older hardware like this typically did not include advanced diagnostic modes.

Basically, Google says it doesn't know and makes provides a suggestion.

[-]

DavidXGA@reddit

Well, let this be a lesson to you and other that AI's don't know when they're wrong, and make up stuff to sound confident.

Don't use them for anything important.

[-]

gnntech@reddit

Absolutely the right answer.

[-]

the_humeister@reddit

AI told me that John Nash (of Nash equilibrium) worked for Microsoft

[-]

aakaase@reddit

💯

[-]

hamburgler26@reddit

Once again seeing this thing I'm blown away at the quality of the screen!

[-]

tes_kitty@reddit

That's normal for a monochrome CRT if everything is adjusted properly.

[-]

Dense_Occasion9971@reddit (OP)

It's remarkably crisp. You even have a choice between white letters on a black background or black letters on a white background with a press of a button.

[-]