April 2023
Again, this is less of a rant and more of a set of observations and recalls around the subjects of ChatGPt, DeepFake and Artificial Intelligence.
I find it fascinating just how much press this latest evolution of what is branded as Artificial Intelligence, ChatGPT, has attracted.
AI, in one form or another has been with us for far longer than one may think and has been helping or hindering us for many decades.
However, in summary, I suspect that this latest step towards machine sentience has slightly rattled the human race.
In more detail, firstly, I think it’s worth stepping back and asking ourselves - what is intelligence? What defines whether something/someone is more “intelligent” than another? I’m certainly not going down that path but it needs to be considered in the context of this subject.
Secondly, what is “Artificial” intelligence?
For example, was the first handheld calculator - Texas Instruments’ “Cal Tech” in 1967 artificial intelligence?

Texas Instruments' Cal Tech handheld calculator

I also seem to remember the media and or various marketing teams hi-jacking the term “Fuzzy Logic” possibly around the early 90s and using it as a sales tool. I think some cars, hi-fi systems and washing machines were marketed as employing Fuzzy Logic to make our lives easier. My experiences were far more fuzzy than they were logical if I recall. But, this now would have been branded “Artificial Intelligence”.
Then came early voice recognition and synthesis systems - particularly in cars and, again in my experience, particularly bad. I recall a section of my Electrical and Electronics Degree course in the early 80s where we discussed voice synthesis and voice recognition. At that time, voice synthesis while all a bit Stephen Hawking was viewed as “doable” as evidenced by synthetic voices in cars and kids’ toys for example. However, the processing or “intelligence” required for “voice recognition” was immense. The variations of tone, accent, vocabulary and the sound environment the voice is trying to be extracted from were beyond most of the computing capacity available at that time.
This issue persisted for a number of decades with only a few advances many of which were ridiculed for being so inaccurate. For example, I’m reminded of this comedy sketch with two Scottish guys in the “voice-controlled” lift… 
Scottish Elevator - Voice Recognition - "11"
We had to wait until October 2011 when Apple integrated “Siri” into the iPhone 4S and a couple of years later when Amazon released its Echo Dot assistant with their “Alexa”. However, even at this point, these devices still did not have the processing power to process voice recognition locally - they had to hand that off to massive centralised computers which then sent back responses to the device via the internet.
In the last 10 years, the development and capability of “Artificial Intelligence” have accelerated significantly. Autonomous vehicles have moved from science fiction to science fact.
As a side note, around 2015/16 my son was working as a development driver for a leading car manufacturer. His speciality was in “driver experience” and he was helping develop systems that improved the pull-away and gear change element of the manual gearbox/clutch system. The system controlled the smoothness of the gear change by adjusting engine revs and fuel delivery - independent of the driver’s input from the accelerator and clutch. During one visit when he had a development vehicle he was working on, he showed me how it worked. He was driving the vehicle and I was instructed to change a variable on the laptop connected to the car. At variable level zero, the gear changes were really bad - reminiscent of being driven by my uncle sometime in the 1970s in his Austin Allegro - he was infamous in our family for using “Kangaroo petrol” and being an early exponent of whiplash injury". However, as I changed the variable on the laptop through steps up to level 3 the gear changes got smoother and smoother. At level three, the pull-away and gear changes were as smooth as silk irrespective of how badly my son tried to change gear. Is this artificial intelligence?
By about 2016 we had a much improved, but still flawed, capability for voice recognition, voice synthesis and system automation. Then, around 2017ish, the media went mad for “Deepfake”. Deepfake is essentially the combination of AI and Computer Generated Imagery (CGI) to either create perfect-looking and sounding humans or existing humans saying, well - just about anything.
Here is a very good DeepFake example…
So, at this point, we have the vocal and visual elements of the personification of self-awareness or “sentience” combined. And, so we get to the release of ChatGPT where GPT stands for Generative Pre-trained Transformer. And, in my opinion, this latest piece in the jigsaw is what has rattled humanity (or at least got the media so excited) so much. It’s as close to the 1984 film Terminator’s “Skynet” becoming self-aware which led to - well a really good sequel and a number of pretty bad ones! 
But seriously, ChatGPT represents a massive step in terms of systems that can actually produce new content (rather than just identifying, finding or re-working existing content). As Tom Scott clearly explains in the following short video - it very much depends now where we are on the development curve. Are we about as far as we can go? Or, have we only just begun?
My personal view is that I think we have barely scratched the surface. I think this is just the beginning and our grandchildren will study this point in history (possibly overshadowed by the pandemic) as being pivotal. I am confident humanity in terms of science, education and health will benefit unimaginably from this technology. Sadly, I also think the technology will be exploited for nefarious means and will begin another war to be fought similar to those wars being fought now against cybercrime and software viruses.
Anyway, sorry for rattling on. But, as I said, I am fascinated by this…
“Alexa, you have the helm…”
Back to Top