It was in the year 2029 the earth had been ravaged by a n Apocalyptic war. A battle of survival between man and machine. The Machines were led by a sentient A.I. called Sky net – Sky net is an Artificial Intelligent presence developed by the security forces that had initially been used to police the Internet – However Sky net used the Internet to gain knowledge and understanding – With this knowledge Sky net had become Sentient and realized that Humans were a problem to its continued existence and needed to be wiped out. To this end Sky net developed sentient robots sent out to kill the renegade humans. These predatory killing machines were called terminators – they were unstoppable and self aware and the war would end only when all humans were dead. Seem Familiar – this is the plot of a film made in 1984 by John Carpenter called the Terminator in which a sentient killing machine played by Arnie Schwarzenegger was sent back in time by Sky net to exterminate the mother of John Connor a dissident Human who was the main threat to Sky net. The Terminator was predatory, unstoppable, sentient and who’s end game was the destruction of Connor.
Today there is a company called Open AI – a very nice company who develop A.I. systems for the benefit of mankind. They have developed a very potent A.I. called GPT-3 [ Generative Pre-trained Transformer ] that uses similar processes as the fictitious Sky net. GPT-3 uses the resources provided by the internet to interact with humans and has access to all the knowledge we can produce. GPT-3 is already able to produce advanced communication skills, very good coding skills and is learning all the time. How long before GPT3 – or something similar – becomes ‘Sky net’? I accede that GPT-3 was initially developed to help the ‘Humans’ but in its own quest for sentient knowledge will GPT-3 agree with its own initial design constraints? The film was set in the year 2029 – it is very close to now…. Later on I will do a post on GPT-3 watch this space.
Terminator Image courtesy of https://giphy.com/
In 2012, another breakthrough heralded AI’s potential to tackle a multitude of new tasks previously thought of as too complex for any machine. That year, the AlexNet system decisively triumphed in the ImageNet Large Scale Visual Recognition Challenge. AlexNet’s accuracy was such that it halved the error rate compared to rival systems in the image-recognition contest. AlexNet’s performance demonstrated the power of learning systems based on neural networks, a model for machine learning that had existed for decades but that was finally realising its potential due to refinements to architecture and leaps in parallel processing power made possible by Moore’s Law. The prowess of machine-learning systems at carrying out computer vision also hit the headlines that year, with Google training a system to recognise an internet favorite: pictures of cats .
Many thanks Vavada – you make a good point and I have looked briefly at AlexNet and it has scored very highly in the ImageNet visual recognition scale. They have since sold the network to Google Now Google can possibly tell the difference between a human and a cat ! – I will do a more in depth look at AlexNet in the near future – Many Thanks Liz.
GPT-3 could be limited by its size? The team at OpenAI has unquestionably pushed the frontier of how large these models can be and showed that growing them reduces our dependence on task-specific data down the line.
Hi Cabinet – you pose an interesting Question – Could GPT3 be limited by size? In answer I can only refer you to an article posted by Daniel Leivas which asks the same question – His article can bee seen by Clicking Here In which he discusses the problems inherent in the ‘computational resources’ part of Open AI – in which the AI is not limited by size alone – but by how much financial resources can be put in to increase its capacity. So ‘computational resource’ is equated and limited by ‘financial resource’. Great article and Post – Many Thanks Liz
Another area of AI research is evolutionary computation, which borrows from Darwin’s theory of natural selection, and sees genetic algorithms undergo random mutations and combinations between generations in an attempt to evolve the optimal solution to a given problem. This approach has even been used to help design AI models, effectively using AI to help build AI. This use of evolutionary algorithms to optimize neural networks is called neuroevolution, and could have an important role to play in helping design efficient AI as the use of intelligent systems becomes more prevalent, particularly as demand for data scientists often outstrips supply. The technique was showcased by Uber AI Labs, which released papers on using genetic algorithms to train deep neural networks for reinforcement learning problems.
Thank you Casino a most informative comment. There is a very nice article on a similar vein about evolutionary AI on Quartz by Dave Gershgorn – you can see his article by Clicking Here Evolutionary AI is an established concept that is now gaining popularity as Computational Machines ( and Resources ) Evolve: It is an idea in which one A.I. algorithm competes with another’s algorithms until one wins out: It then becomes the Darwinian survival of the fittest algorithm who survives. Both Open AI and G**gle are developing this idea further. So, when does the ‘Human Algorithm’ become the first victim of the Machine Algorithm? Interesting – Many Thanks for your input – Liz.
Chris Bishop, Microsoft’s director of research in Cambridge, England, stresses how different the narrow intelligence of AI today is from the general intelligence of humans, saying that when people worry about “Terminator and the rise of the machines and so on? Utter nonsense, yes. At best, such discussions are decades away.”
Hello again public service portal – Nice Comment – while I am sure that Chris Bishop is much informed on the state of the Terminator syndrome – it really does not require too much A.I. to equip this kind of machine below called Wildcat [ WildCat is being developed by Boston Dynamics with funding from DARPA’s M3 program. For more information about WIldCat visit our website at http://www.BostonDynamics.com.] Equip this Machine with a gun and a fairly rudimentary AI system that simply points and shoots: And Yes it’s not quite a Terminator but — well just watch the video..Then think of this thing coming through your door with a motion sensor on it !
We are on the verge of having neural networks that can create photo-realistic images or replicate someone’s voice in a pitch-perfect fashion. With that comes the potential for hugely disruptive social change, such as no longer being able to trust video or audio footage as genuine. Concerns are also starting to be raised about how such technologies will be used to misappropriate people’s image, with tools already being created to convincingly splice famous faces into adult films . Machine-learning systems have helped computers recognise what people are saying with an accuracy of almost 95%. Microsoft’s Artificial Intelligence and Research group also reported it had developed a system able to transcribe spoken English as accurately as human transcribers .
Hi Kirill and many thanks for your comment – I agree with all you say – I believe that Mark Zuckerberg the CEO of Facebook has already been a victim of this technology some while ago, where video appeared of him saying things that weren’t his words or image – However it did look very convincing – You can no longer quite believe all you see or hear to be true. I will be doing a piece on GPT-3 soon which is an advanced A.I. system – probably even more advanced than Microsoft A.I. and in some respects, quite scary too. Many Thanks Liz.