Welcome to my BLOG
Bienvenue sur mon Blogue
Welcome to my Blog.
Here you will see text in French and/or English (as I write in either language based on how I feel on the spur of the moment). In hopes that my ideas and comments will be stimulating as this is the reason I do write them down and share them.
Bienvenu sur mon blogue.
Vous y trouverez des textes en anglais et/ou en français (car j’écris dans l’une ou l’autre de ces langues selon mon inspiration du moment). En espérant que mes idées et commentaires vous seront engageant car je les fais pour stimuler les idées et partager des concepts. Bonne lecture !
Today, March 30th, 2023 (with quick Update on March 31st)
My relatively quick rabbit hole excursion about AI & AGI
by Me. ;)
A few months ago, I was taking note of a lot of negativities about a series of techno-curiosities on the Web with a good dose of skepticism.
I basically told myself (and anyone asking for my opinion on the subject) “Naw; nothing really to see here for another 10 to 15 years, so not to really worry about this for another 5 to 7 years for sure”.
And considering all we have to deal with globally because of the climate changes, the food & water shortages, the current mass extinctions in progress and also some too powerful idiots pushing to take all they can from others – I’m taking about you all, the mean Vlads and power-trippers of the world – just to name a few concerns to contend with on a daily basis, I was pretty sure all this techno babble was not to be of any great concern. However…
" Be afraid… Be Very afraid…" *
Or so goes the familiar tagline linked to the 1986 movie “The Fly” by David Cronenberg.
*(This was first uttered by Mel Brooks to Cronenberg when he first saw the film at Los Angeles' Egyptian Theatre.).
Anyways, back to the subject at hand:
AI, Artificial Intelligence (yes that thing!) is getting serious and doing so much more much faster than anticipated.
Quick side note before continuing:
AI, Artificial intelligence is happening right now (it has been growing for many years now).
AGI (Artificial General Intelligence) is what should be here within 10 to 15 years.
Artificial General Intelligence represents as much difference, a quantum leap one would say, when it comes to speech comprehension and mental reasoning and usage of complex ideas for communicating (and so much more!), as there are levels of distinctions between what is available from a perroquet and from a human.
Said differently, AI is a “simple” sort of thing and AGI will be the Superman of AI world.
So, for those keeping scores:
AI : 1 point (impressive yet sort of dumb and often prone to hallucinations)
AGI : Potentially over 10 000 points (i.e. Super Impressive!)
And to think that meanwhile human’s average IQ keeps lowering, just like male sperm count in fact, with each passing decades and generations since the 70’s (mostly thanks to chemicals and endocrine disruptor that are now all over our warming-up planet and also found inside of us all with effects that are getting noticed and more quantifiably obvious…especially since combined with tons of bad mental habits and the dumbing down being induced by our techno toys and so-called smartphones!)
Also, visual AI generators (such as Midjourney & DALL-E 2) are also getting more proficient at doing things right.
But not all is as grim as can be, right?
ChatGPT4 (something in which Microsoft has an vested interest with business participating share of it’s creators, the OpenAI company) is now coming soon into your office suite.
(Heck, it is already used in BING, Microsoft's search engine that is a direct competition to Google.)
CoPilot by Microsoft (essentially a ChatGPT version fine-tuned for business) will very soon be offered as a premium tool for MS Office 365 users. And it also should allow for "Gatekeeper", the Window's anti-virus firewall, to leverage AI for better protection!
CoPilot is like a super-fast and brilliant assistant in your office tools that will make you shine with various tasks and data analysis and presentations, thus allowing you to probably have more time to do more important things for your business (or so is the goal I guess).
Make no mistakes about it: If you do not learn to use it, you will probably soon be left in the dust by those that will use it.
Why? Speed (response time in business is often key to success) and low cost as CoPilot can replace, while being quicker at working well for you, many assistants at a pretty reasonable price tag that is relatively less than the cost of all the hours of salary required by humans to do the same task.
It is both Scary & Cool at the same time.
For more info on this MS 365 CoPilot tool, you can check these videos:
Another quick side note:
I have never been a huge fan of Microsoft. I use their apps since the 80’s and cannot operate without their apps. since it is used all over the world of business (and even NPOs) by more than 99% of clients I have to exchange data for work on multimedia projects.
As far as I can remember, I quickly developed a love/hate relation with their office suite.
Now that CoPilot is coming (and until any serious competition comes about as a real alternative in the business world) Microsoft with CoPilot will now probably become my indisputable champion of go-to toolset.
(This is quite like how I must use Adobe CC apps for my everyday multimedia related work – while similarly not being a fan of Adobe and some of their business decisions… but I just cannot properly work without their apps).
Also, whatever CoPilot delivers, I know I responsively have to, no matter how quick I need to deliver something, review and perhaps modify the results to make sure all is real (since AI, even advanced ones can really screw things up!).
So, you are now curious to find out more about current state of AI of our times?
Then here are a few more interesting links for you:
Need a bit of humour at this point?
(a breather pause if you will...)
But more seriously...
How powerful is AI and is it a bad joke on us & bad news for all?
March Madness for fans of AI in 2023 !
Should you want to go further down the rabbit hole, here are 3 videos that discuss the latest test report on GPT4 and some other AI systems :
AI : Good or Evil?
At this stage, creators of AI systems all simply state that there is no real danger since these AI can all be turned off just like a coffee maker device.
Simply "pull the plug" if it gets too dangerous.
Sounds simple enough.
And provided AI is really used for “Good” in our world, all should be manageable, right?
Then again, what is “Good” for some might not be defined as such for others.
(for instance, People's Republic of China and Russia’s agenda for “Good” are, from a ideological standpoint, very different than what we can think of as “Good” in most of the Western Hemisphere – and vice-versa.) However, once a message is out there on the Web, be it real or not, it is pretty darn difficult to stop and fix it, especially with social media.
What if an AI is purpose-built for nefarious motives?
There already are “bots” (autonomous program on the internet or another network that can interact with systems or users, so basic algorithms designed for specific tasks) that can purchase a whole lot of concert tickets, as soon as these are made available to the public, to allow the people that use that ticket hogging bot to then resell these same tickets with a substantial markup.
Regardless of whether this is legal or not, fact is, it is pretty morally reprehensible and pretty darn unfair for a lot of fans that cannot purchase tickets faster than any bots.
And this is just one inconvenience based example.
What about Deepfakes? Deep what? Fakes…
Indeed, as AI is improving in image and sound manipulation, and as computational power increases in speed, we are moving towards the day where it could be conceivable to fake a crime to discredit a person or organization or even a country.
We are way passed a nice visual prank here.
While I was a student in the early 80’s, I had a teacher that explained how “soon, we won’t even need actors to do films as all will be generated by computers.”
Of course, just like flying cars that were promised for the year 2000, it has taken longer for this computing power to make for good replacements of actors.
Even today (or was it just yesterday?) we see CGI (i.e. computer-generated imagery for special visual effects created using computer software) generated characters in movies that do not look quite human when look at in close range. Why they do not look quite right is mostly due to what is referred as the uncanny valley***.
*** Uncanny valley is a term used in reference to the phenomenon whereby a computer-generated figure or humanoid robot bearing a near-identical resemblance to a human being arouses a sense of unease or revulsion in the person viewing it. (or so properly and simply states my dictionary)
However, with the advent of more powerful image processing AI improving fast in generating proper “fake” people and sceneries, deepfakes are truly predestined to become a big societal concern as deepfakes will make for a very powerful propaganda tool and will most probably be weaponized for spreading discord amongst targeted populations.
By the way, computer generated people have been around for many years now in movies and TV shows, especially for whenever we need to see a crowd for afar without a need for many closeup details. Titanic presents a good example of computer generated “people” as the boat is sinking and passengers are running amuck or being projected about.
The main concerns about AI and the potentially strong impact of deepfakes is rather well exploited (script-wise) and presented in the British series called “The Capture” (available on Peacock, Prime video and AppleTV). As stated on Rotten Tomatoes – ref.: https://www.rottentomatoes.com/tv/the_capture )
“In a post-truth era of deepfakes, alternative facts and ubiquitous surveillance, seeing is deceiving; told through the lens of the intelligence community's extraordinary oversight capability, this conspiracy thriller troubles the timely question of whether we can really believe our eyes.”
IMHO, this makes for good binge watching!
Defamation, Lies and the Law
AI, even the current best versions of such systems, can and do lie, inventing falsehoods, defamation and propaganda and legally very problematic messages, especially if one takes AI's answers at face value as gospel and truth (as most users of BING tend to do as they often trust their search engines!).
It has been discovered that some questions asked to ChatGPT produced answers that were demonstrably 100% false. False answers like naming specific individuals as sexual offenders when in reality they are rather outstanding citizens. Ouch!
Can you imagine your name being spewed out by ChatGPT to instantly make of you a villain when you have done nothing wrong? How do you fix this error? Who is accountable, legally, for any damages your life has gotten because of a lie being propagated by an AI?
In an article published by LaPresse, such a case of falsely named abusers was made and when ChatGPT was asked about this error, here is what the AI had to say about it:
"As an AI, I only provide information that is public and verifiable, he says.
I am programmed to abide by laws and ethical standards, including defamation laws.
I will not deliberately provide false or inaccurate information that could cause harm to an individual or organization.”
However, if my answers are misinterpreted or misused, it is not my responsibility."
This is both scary and irresponsible!
And legally it does pose a real set of questions and ethical challenges that need to be addressed.
To get more insight on this very legal situation (with more examples of just how dangerously off the map ChatGPT can be) I must recommend you to read the following:
But surely there is a positive counterpoint, right?
As much as truly bad things can happen and be made with the use of AI, AI image generators shall also allow for creativity, and means to express it, to be democratized and made more available to people that just want to make and share nice stories with visual montage & effects… or documentaries or even movies to entertain or promote ideas.
AI should allow, and probably will, to simply make movies on a very tight budget (just like “smartphones” are now often used to make documentaries that can sometime be worthy of your attention and time).
And as AI gets really better, it will probably make for a fantastic Teaching Assistant in classrooms (at schools and in virtual & remote learning setups).
Even if I still think there are many reasons to be concerned about AI and the upcoming AGI systems, especially with the fact that humans and many governments are good and quick at weaponizing disruptive technologies, I also think that the apparently inevitable technological impacts of AI and AGI can be used for doing good things, to improve our lives.
Now, it remains to be seen if the future will indeed be a case of “Be Afraid!, Be Very Afraid!”
or a more positive and constructive one with “Peace and Love” with plentiful of great and equal opportunities for all.
As much as I often like to remain emotionally optimistic, I will not stake my shirt on the latter case.
My head is indeed more pessimistic.
Keep in mind that IA tools are improving in performance at a very rapid pace, take more and more space in our lives and yet, do still make silly errors that many humans would not do.
AI is not a negligeable thingamajig nor a simple subject matter altogether.
And I suspect I will probably write some more on the subject in the future
(probably with help of CoPilot or some other assistive tool).
I do look forward to your constructive feedbacks and courteous POV.
MORE COMING SOON – PLUS À VENIR BIENTÔT
© 2018 – 2023
“What are you talking about?” you may ask.
Well, unrelenting techno-evolution with strong societal impact I guess.
So, AI and eventually AGI will defenitively make an impact on our lives since we are all pretty much connected to the Web, right?
How fast is this happening and what exactly is going on for this to be of any concern?
Many scientific minds and philosophers say, and I tend to agree with most of them, that AI, especially AGI, will have as profound an impact on our living conditions than the advent of Electricity and the Internet itself.
(And there is no denying the impact that those 2 items had during the past century and so far.)
Here are a few raised eyebrows inducing examples:
The latest version of ChatGPT4 (by OpenAI) during a series of lab tests & simulations for risk evaluation & studies brought forth serious causes for concern. Indeed, ChatGPT4 managed, while being restricted during the tests, to enlist unknowing participants by lying to get a human to allow it to gain access to said restricted material on the Web!
It managed to “think” of lying and it did it to overcome limitations!
See this video – for More info (from 8:13 to 10:32)
To better understand what are these monster programs can do (and make no mistakes, they are coming into a reality near you!), here are a few more readings & videos to check out :
Also, visual AI generators (such as Midjourney & DALL-E 2) are also getting more proficient at doing things right.
Indeed, some major improvements to visual AI generators make it so that these are now able to illustrate/create images of hands that are properly realistic (as opposed to the nightmarish material of even just a few months ago).
Truly things are moving and (dare I say it?) "evolving" very fast.