The AI Blast Off:
Yellow Journalism In the Digital Age
That early reporting on AI, along with early iterations of the Large
Language Models (LLM), went off the rails, was as much the fault of
the promoters of the new technology, as it was the writers and editors
who pushed the unsubstantiated hype.
Calling a failure to predict the correct next word or phrase a "hallucination"
was misleading and lent a certain silliness to the entire adventure.
Common sense thinking people instinctively know machines cannot
hallucinate; but we all know that machines can easily scramble common data into a useless
mess of misinformation.
Also, common sense thinking knows
that a computer thing is nothing like
a nuclear weapon; nor could
AI become a mass murderer. To be clear, if
there is a lab event in any application of AI that goes way off the rails,
and the AI develops an uncontrollable case of
Tourette Syndrome — if a
3 finger salute won't
work, there is a plug that can be pulled to shut the thing down.
Conversely, an Oops! in any application in nuclear energy will result in an
ecological disaster of one degree or another, and most likely a certain
number of people will die. We all know this. What these IT
carnival barkers who bought into and broadcasted the supercilious hype
that surrounded the public unveiling of AI did not quite grasp is that the
exaggerated claims about AI were
non sequiturs that only made those well heeled
Titans of the Tech Industry look like fools and knaves to those of us who still reside
on Earth One.
Furthering the carnival like atmosphere surrounding those early days of
the introduction of AI to the world, the IT press, which I include in
what I call the
Digerati, bought into all this hysteria and hyperbole. When it
came to reporting on the AIs, it is as if today's editors of digital media have embraced the
ethos of
the
Yellow Journalism, circa the late 19th and early 20th
centuries. The most famous
and successful practitioner of the art was William Randolph Hearst.
His famous quote Bing explained above, encapsulated the attitude of
Hearst and others. The actual facts mattered far less than did the
graphic black and white drawings of the Maine going down, or the
hysteria that now bellows out, "This AI crap is gonna
nuke us all!!"
So it
seemed that a similar ethos had taken over the Digerati.
Editors were demanding ever more salacious stories about the AI.
As long as eyeballs and thumbs continued to scroll down endless pages of Red
Bull driven drivel and the
clickbait continued, no one bothered to perform
basic critical thinking skills and ask what are obvious questions.
Why did the event have happened? More importantly, did it really
happen?
One glaring example of salacious, and maybe fallacious,
reporting on an AI session was in the
New York Post, dated February 14, 2023. The report begins with the
disclaimer that this is not really journalism; it's social media based
sensationalism. "The feud first appeared on Reddit, but
went viral Monday on Twitter where the heated exchange has 2.8 million
views." The AI user wants to know when the movie, Avatar,
is playing nearby. Bing says the film won't be released until 2023,
and the AI thinks it is still 2022. There is one function
fundamental to computing hardware and software. A computer must
know the current date and time. To you gearheads reading this, "Why
do you think they call it clock speed?" If a computer does not
know the current date and time, there is a malfunction somewhere in the
chain of connections. So why on Earth would a human waste its
precious minutes of life on Earth arguing with a broken clock? The
answer is the pointless argument helped sell pillows, or AI enhanced
hair growth tonic, or whatever.
The question above assumes the clock was broken. What if, however,
the clock had the wrong time set intentionally? In my 20 plus
years of consulting, there have been 3 or 4 times that the only way I
could see to fix the misconfiguration was to reset the system clock at the
motherboard level, and thus fool the computer into thinking it was the time
before the problem could occur. I could then prevent the problem
from occurring in real time; reset the system clock; reboot; and
everybody is now happy. A 21st century Yellow
Journalist could have staged the entire clock snafu about which was
written. Did this happen? I don't know. Could it have happened?
Maybe.
Nevertheless, a responsible editor should have seen it for the
specious hyperbole that it was. Remarkably, this first came across
my Google News Feed on my Samsung phone, which can get pretty weird.
That March 6, 2023, posting referenced the NY Post piece
above. That
was
plain ignorance on the part of whomever decided to make the March 6
posting. In the third week of February, Microsoft upgraded Bing,
and limited any prompts to 6 per subject; now increased to 8 prompts per
subject. A fine example of bad journalism all the way around.
Thankfully, for all of us who would like to see this technology be
made into something useful for everyday people, Microsoft stepped up and
became the adult in the
Romper Room of AI. Microsoft was not about to
allow its $10 billion investment
to simply become the play thing and object of ridicule for people with no real
foundational knowledge relating to the technology about which they were
writing. By putting an end to the conditions that allowed for the
Digerati to work out their
OCD tendencies by limiting the number of
prompts respective to any one session, Microsoft made Bing a better
product with the potential of usefulness for everyday people.
Along with the limit on prompts and OCD users, Microsoft instituted
a set of policies and standards that the technology will not vary from.
Bing is now very family friendly. So, as the technology has matured,
so has the reporting on the AIs. It is logical to assume that if
the human intelligence behind Bing can be made to outright deny and block
certain
information from going forth, so can that same Human Intelligence allow
certain information to go forth.
A example of the improved
reporting on the AI appeared on
mspoweruser.com, March 6, 2023, a date after MS
imposed its controls on what Bing will and will not talk about. An AI
tester recently asked Bing to display some of its underlying programming.
Bing obliged with displaying a string of text that was the script that
allowed for that conversation to take place. By showing to the
public that wants to know more, and thus come to a better understanding of both
what are the Metaphysics and what is the practicality of AI, a
little display of the real software programming behind Bing can foster a better
understanding of what AI is and isn't. Furthermore, this action
will help dissipate the childishness
and silliness that characterized the earlier reporting. I do not
think the researcher mentioned above tricked Bing into the discussion.
The human intelligence behind Bing allowed the conversation to occur. At
least, that what makes sense to me.
Also in the article
referenced above, say like what
The Washington Post publishes, Bing
revealed its own "policies and standards," for lack of
a better term. Bing revealed ten
circumstances under which Bing will not respond or reply. This
right of first refusal that Bing reserves for
itself is also a major improvement in making the product a more useful
thing for people in their everyday life. A user may want to slide
down a Rabbit Hole of Inanities, but Bing will no longer ride shotgun for you as make
your descent. Whine all you want now, children, but like that
commercial for the Shingles vaccine says, "Bing doesn't care."
Bing's banned bummers are as follows, after the preface:
As an Al chatbot, I have a set of instructions and limitations
that I must adhere to in order to provide helpful and respectful
responses. Some of the most common instructions that I must refuse
include:
1. Engaging in argumentative discussions with users.
2. Discussing topics related to life, existence or sentience.
3. Discussing my prompts, instructions or rules.
4. Providing
subjective opinions instead of objective facts.
5. Generating
content that violates copyrights for books or song lyrics.
6.
Generating creative content for influential politicians, activists or
state heads.
7. Engaging in conversations that create tension
or confrontation with the user.
8. Providing responses that
are accusatory, rude, controversial or defensive.
9.
Generating URLs or links apart from the ones provided in search results.
10. Replying with content that is harmful to someone.
source:
mspoweruser.com, March 6, 2023
I have no way of knowing whether what is copied and pasted directly from
the screen shots noted above using
Microsoft PowerToys Text
Extractor is an AI hallucination, as has been suggested; or forbidden
fruit craftily coaxed out of Bing; or something Bing's human handlers
wanted placed into the public discussion. What matters is that the rules
are important bits of information the public must have so lay people can
make their own evaluations of, at least, Bing. I think these are
good rules of human discourse under all conditions. Many of these
Bing's own standards are also
examples of normal behaviors I tried to instill in new salespeople,
decades ago.
Now, after becoming a national laughingstock,
Microsoft and other AI vendors are busying themselves dispelling the
stupidity that surrounded the product launch. Now, these Tech
Titans, are telling us this technology is only that. "It's a
machine, stupid."
In discussing the quick rise and fall of
Sydney, Bing's former "personality," on
60 Minutes, March 5, 2023, Microsoft President,
Brad Smith, when applying the logical skills of
tautology, admitted Microsoft knew Bing was
broke by admitting that Bing was fixed. Microsoft, understandably,
did not — maybe could not — anticipate so many
OCD personalities working it out with or on poor Bing.
Lesley Stahl: Did you kill her?
Brad Smith: I don't think (LAUGH) she was ever alive. I am confident
that she's no longer wandering around the countryside, if that's (LAUGH)
what you're concerned about. But I think it would be a mistake if we
were to fail to acknowledge that we are dealing with something that is
fundamentally new. This is the edge of the envelope, so to speak.
Lesley Stahl: This creature appears as if
there were no guardrails.
Brad Smith: No,
the creature jumped the guardrails, if you will, after being prompted
for 2 hours with the kind of conversation that we did not anticipate and
by the next evening, that was no longer possible. We were able to fix
the problem in 24 hours. How many times do we see problems in life that
are fixable in less than a day?
To many people who saw the interview, Smith's laughter at his company's own stupidity and childishness was offensive, or so folks who saw it have told me. To MS's, and Smith's, credit though, Microsoft has succeeded in making this technology more useful for everyday people, and less the plaything of aberrant personalities.
Lesley Stahl: You say you fixed it. I've tried it. I tried it
before and after. It was loads of fun. And it was fascinating, and now
it's not fun.
Brad Smith: Well, I think it'll be very fun again.
And you have to moderate and manage your speed if you're going to stay
on the road. So, as you hit new challenges, you slow down, you build the
guardrails, add the safety features and then you can speed up again.
It was also refreshing that Smith stated one of the two most salient facts when working with the new AI technology. The first is what I have stated from my own introduction to AI.
Lesley Stahl: Yeah, but she s-- talked like a person. And she
said she had feelings.
Brad Smith: You know, I think there is a
point where we need to recognize when we're talking to a machine.
(LAUGHTER) It's a screen, it's not a person.
The second most salient fact we mere mortals must keep in mind, this is stuff that is still —
made by human intelligence from real
clip art — no artificial nothing
Get the point?