Will HAL (your AI wing-man) drive you to Centrelink?
It has been a busy year, the pace of change keeps accelerating, the need to run always faster just to keep up is absolute. Therefore, the pace of failure is also accelerating, the lives of businesses getting progressively shorter, from the corner store to massive multinationals. They emerge, flower, then disappear, either by being taken over in some way, or just doing an ‘FTX‘ and exploding in spectacular fashion after a life counted in months.
In November, the newest, shiniest thing ever, hit the spotlight.
The all singing, all dancing AI system that like all previous new shiny things will change everything. The hype is enormous, and if only some of the promises made are fulfilled, it is truly a step-change, but we have heard it all before.
For those not tuned in, ChatGPT is to the ‘chatbot’ we seem to be meeting whenever we want to communicate with an insurance company, bank, or internet provider, as a go-cart is to a formula 1 car. A sailing dinghy to a 12 meter foil catamaran.
This thing can keep an every-day conversation going, write books, code, explain quantum mechanics to Einstein, beat a line-up of Go grandmasters while sleeping, and the scribblers of blog posts like me, are now utterly redundant. In short, it communicates in natural language, with natural and personalised nuances, backed up by the resources of the web, integrated into a conversation in real time.
Absolutely liberating, or absolutely scary, take your pick.
HAL (Heuristically programmed Algorithmic computer) the product of Arthur C. Clarke’s vivid and futuristic imagination, and star of the 1968 film ‘2001: A Space Odyssey’, is here.
Maybe.
Assuming at least some of the hype is real, and it certainly seems to be, we have another round of introspection about the role of humans in our commercial and creative world about to become a serious conversation. Get ready for it in 2023.
UPdate Dec 27th.
The hype is real. HAL has come home, and seems to be taking charge.
Yesterday afternoon, accompanies by a few post Xmas sherbets, a couple of us (me a business strategist, a scientist, a specialist doctor, and a writer), did some Chat-wrangling to test out the system. Across a range of domains, we were blown away with the capabilities. It is clear that teaching institutions suddenly have a huge challenge to face, as this thing can produce in a few minutes a highly credible response to complex challenges, including references used. That is only the start to the downsides, and while the eminent group of post Xmas testers did begin to see a few upsides, on balance, ChatGPT has opened a door previously locked tightly shut.
Post script January 10, 2023. Chat GPT is only one of the AI tools coming at us. The speed of innovation is way beyoind the capability of regulators to even understand this stuff, let alone regulate the output. This is an informed conversation on ‘Regenerative AI’ and what it is bringing to the table. If you are tbinking about the implications, it is worth a look.
Second Post script January 15, 2023. Commentators everywhere are waking up to the challenge suddenly in front of our education system and educators. ChatGPT has removed the easy ‘tick and flick’ essay marking, from primary school to post graduate levels. As with all technology, it will only become more sophisticated. A final exam in my degree, 50 years ago, was a Q&A exercise. I had plenty of notice of the format, conducted in the home of the professor where he had a wide ‘Gone in the Wind’ type staircase. He started with some questions, and we had a conversation, all the time him indicating I should back up the staircase, or come down. To pass, I had to get to the floor, to fail, I only had to reach the top of the stairs. Eventually, after a nerve wracking 40 minutes or so I reached the floor, and he gave me a congratulatory whack on the back, and a sherry. (hate sherry). The knowledge that this was the format had forced me to think deeply about the subject, and read widely, as well as engaging the Professor several time during the semester to clarify the ambiguities present. At the time I recognised the value, but it was unusual to say the least, and an entirely different approach to teaching. This post from Seth Godin on the way forward for teaching, implies my professor was only 50 years ahead of his time, and me, a guinea pig.
Third Post script Jan 18, 2023. it just keep ‘getting better’ as more is understood about this tool. This article from a medical journal has ChatGPT passing the exams to become a doctor. This has implications both good and bad. For example, the ability to access data base records of bone trauma, malignant growth, could make diagnosis logarithmically more accurate, but I am not so sure about that rectal exam.
These tools have also created a greater opportunity for cybercriminals to get into our collective pockets. This is a truly scary development.
Fourth Post script. Feb 19, 2023. ‘Where to from here’ is the question everyone is asking themselves, and others. I am sure it is keeping Google executives up at night as they see their Golden Goose cash machine strangled by a Chatbot enabled Bing, and probably others finding a crack in the fortress wall.
This post ‘when the internet becomes Chat- People vs Algorithms’ takes the best shot at answering that question I have seen. Hat tip to Mitch Joel who pointed it out to me.
This second post I found, also on Feb 19 explains the workings of ChatGPT, and presumbaly its emerging competiotors. We need to understand this, as the technology is being integrated into our lives at a rate logarithmically faster than anything I have never seen before. How ChatGPT Works: The Model Behind The Bot – Towards Data Science
Fifth Post script Feb 26, 2023. This article courtesy of Stephen Wolfram is a long, but detailed laymans descciption of the workings of ChatGPT, and all other ‘Large Language Models’ that are around, and in development. It is well worth your time, assuming you are one of those who is motivated to understand how stuff works. I am sure there will be many more, between them we digital novices (*like me) might gather enough understanding to make sense when we pontificate.
Sixth Post script March 19, 2023. ChatGPT4 was released to the world on Monday (March 13, 2023). The blurb tells us it is a step change ahead of the previous version, which we are only just starting to come to grips with.
Open AI is as the name implies, open source. Therefore there has been a tsunami of apps coming at us that do stuff that was impossible a year ago, and that tsunami is a ripple compared to what is coming.
Take for example the ability to generate an AI video during a coffee break with no more skill than most of us have already. Synthesia.io can do it for us. The utility of this capability is enormous. It offers the opportunity to save time and money while delivering everything from a public presentation to internal training and induction videos, email list ‘welcomes’, and product demos, better than anything possible until now.
In this short video Seth Godin displays an AI capability that enables characters in a movie to say what you want them to say, and have their lips and movements be exactly synchronised. I understand this is currently available to ‘Hollywood’ producers for bags of money, but inevitably will be open to the rest of us for a few dollars in coming weeks.
The list goes on, and will only grow exponentially as some of the most creative people in the world set their minds to leveraging this stuff of Arthur C. Clarkes imagination.
Seventh Post script. March 25, 2023. The ‘Chatbot’ war is joined. OpenAI has launched ChatGPT4 as a subscription service, and Google has re-released ‘Bard’ after the clusterf**k that was the first try a few weeks ago. This article examines the differances between ChatGPT3 and 4, as well as having some very useful links that give answers to other question in this space that you may have.
Frankly, If you are not thinking about the exponentially accelerating rate of change being driven by this technology, and the ways you can use it, you will become increasingly unable to compete. Like any investment, it takes an ‘upfront’ payment and some risk to put yourself in a position where you can reap a return.
8Th Post script March 29, 2023. This world of AI is regenerating itself at a compounding speed. Scary stuff, or as exciting as anything that has ever happened? depends on your point of view. This video which summarises the latest AI developments is instructive. Hat-tip to Steve Aspey for pointing me to it.
9th post script. April 4, 2023. A further very useful explanation of how ChatGPT works from Stephen Wolfram. It is becoming clearer by the day that we have reached an inflection point, and there is no going back. Currently there is a debate in the US sponsored by several digital luminaries, calling for a 6 month halt in development, concerned that the tech is running too far ahead of the moral understanding we have. Great idea if everyone (Russia, China, north Korea, et al) agreed, but in the absence of that agreement, not a good idea. Why give them a head start in the race of the decade? A YouTube addition to this PS. Howard Weiner pointed out to me that I missed the Youtube explanations of the way this technology works. I learn visually, so the videos have been remarkably helpful in building understanding. Long way to go yet!
10th Post Script April 6, 2023. This space is evolving with lightning speed as we figure out how large language Models (LLM’s) will impact our lives. What about ‘Knowledge Graphs?. Here is an ;interview researcher Kurt Cagle conducted with Chat GPT4. It is not dystopian, but getting there. It is well worth a read, if nothing else, it should spark some thoughts about how you might be affected.
11th Post Script. April 9, 2023. ‘What is next? seems to be a fair question. Who knows should be a fair answer. One of those who thinks deeply about these things then published his thoughts is Christopher Penn. In this piece, after offering a bit of history, he forecasts that LLM’s will actually get smaller, after reaching a gargantuan size, too big for any but the largest installations to process. This makes sense and happens every time in any evolutionary environment. Things get big, then they progressively get smaller and more focussed, the specialist outperforming the generalist at domain specific tasks.
I think he is right, what do you think?
12th Post script. April 20, 2023 AI comes to imagery. You might expect Canva, and other established players to rapidly incorporate AI into their offerings, and this is happening. In addition, there has been a rash of start-ups and SME’s that have led the goldrush. This twitter thread I stumbled across has some mind-boggling examples of AI driven imagery. I sent it to a mate who is a professional photographer, and he just shook his head, and wondered where the next arrow in his back was going to come from.
13th Post script. April 26, 2023. This is getting really interesting!! The pace of innovation is just astonishing. Two things this morning, the first a finished ad, made entirely with AI. https://tinyurl.com/39nftyvaWhatever you think of the ad, making it entirely from AI opens up extraordinary budget productivity increases for marketers seeking ‘Activation’ ads. Then Steve Aspey sent me this video, which is an A-Z on how to make such an ad.https://www.youtube.com/watch?v=h3AhYJ8YVss.
14th Post script May 1, 2023
Academics are inclined to moan, and set out to ban AI tools in exams, thus protecting their territory. This post from Visual Capitalist shows exactly why exams as we all knew them to date, are now utterly redundant.
My brain is hurting.
15th Post script May 10, 2023.
It is fair to be wondering where this is all headed. Clearly, at least to me, AI will impact every person, and every job on the planet more quickly than most would forecast. This in one path, probably one of many, from someone in a position to know. https://www.youtube.com/watch?v=gMsQO5u7-NQ
16th Post script May 18, 2023
Some may have seen references to Geoffrey Hinton stepping down from his role at Google in order to more freely discuss the nature and role of AI in our lives. Hinton is widely known as the ‘father of AI’. His views are not the noisy, superficial opinions that have sprouted like mushrooms after rain since ChatGPT was launched on the world in November last year. They are the views of a polymath who has been researching this area for 30 years, and who trained one of the founders of OpenAI, the ‘startup’ that created ChatGPT.
This article from ‘The Conversation’ examines Hinton’s view that we should not think about AI as a substitute to our own cognitive ability, but as a different sort of intelligence that should be complementary to our own, seeking solutions to problems from a different perspective than we humans bring.
17th Post script. May 25, 2003.
The creative and productivity upsides of regenerative AI are emerging, about as fast as the downsides. The potential for identity theft enabled by this technology is truly scary. I have been following the path of the AI revolution since I first stumbled across ChatGPT in early December last year. I first asked myself the obvious question for us all: ‘Will this take my job’. Clearly the answer in many cases is Yes. This TED talk conversation shows just how easy it will become to be someone else.
Truly scary stuff, evolving at lightening speed. Our regulators have not got their heads around social media platforms despite the evidence of the necessity in the 15 years since they became ubiquitous. We cannot expect anything different this time, which means the world as we know it has changed into a ‘hunger games’ type competition, with us as the prize.
18th Post script June 10, 2023.
I continue to hear, in almost equal measure, those who give almost spiritual power to AI, and those who catastrophise it as the coming end of the world. I sit firmly on the fence, seeing huge benefit to us all, as well as the downsides when (not if) bad people are able to harness the power of AI for their purposes at the expense of the rest of us. balance is required. This post by Marc Andreesson who has some credibility in the field, is firmly in the ‘Pro AI’ team. it is a cogent argument for AI, and worth your time to read.
On the ‘Con AI’ team is a bunch of equally qualified experts who hold the view that the machines will ‘evolve’ to such a point where they do get us. The dystopian movies are just simple forecasts. Firmly in this camp is Mo Gawdat, the former boss cocky at Google X. In this podcast, he paints a picture that might stop millennials having children.
During the week just past, Industry Minister Ed Husic released a discussion paper for ‘Safe and responsible AI’. I am very glad the government has twigged that there is something going on here, and we need to be discussing it, but the reality is that Australia is just a flea on the dogs tail. The US and China will set the rules of the game, and there is no way in my view that they will be compatible. The US is engaged in a huge new grab for position in an industry that will determine the future, and the cash that goes with it. China is all about control, they will exercise it irrespective of what anyone else thinks. Strap in!!
19th Post script August 16, 2023
Into my inbox comes this long post from Ben Evans, which answers the question I first asked way back in December last year when ChatGPT3 burst onto the scene: Will AI take your job? The answer is ‘Yes, it might, but there will be many more jobs created as a result.
The post details the history of labour saving devices from the introduction of coal powered steam through to this current AI revolution. Every time, the introduction of new technology results in an improvement in living standards and the number and type of jobs that emerge. There is no reason to believe this time it will be any different, it is just that we have great trouble doing the forecasting of where those jobs will be. That job requires hindsight.
20th and last Post script. November 7, 2023.
It is two weeks short of a year since ChatGPT was launched on November 30, 2022. It seems appropriate to mark that anniversary with a final PS on this post. Yesterday Chat announced the launch of GPT’s a set of tools that will enable development of bots without any code. To me, it is the first step in the logical sequence of development of AI into tools that car customised to meet the needs of individual use cases. I am sure others will follow, indeed they may be already there, and i have just not stumbled across them.
In any event, the explosion of AI, and we are only in its early stages, the ‘Model T’ era of AI, has already changed lives, and that pace will only accelerate. Good luck managing that!!
Post Script. January 14, 2024.
I had not meant to add to this post, now over a year old. However, the question of the degree to which AI is just another digital bubble has to be asked, and to ask it, who better than Cory Doctorow?
