Artificial Intelligence Revolution Series Part I – Generative AI: A Super Power in the making

Geoffrey Hinton`s resignation this week from Google should make all worried because Geoffrey Hinton was our guide toward developing AI. He had his PHD in Artificial Intelligence 45 years ago. He has set the path of developing AI through his biological approach. His vision on AI wasn’t possible back then due to the lack of data and capable hardware. Now, that Geoffrey Hinton has achieved this great achievement; Generative AI, he is backing off? He will start to warn of AI dangers? Life has taught us that when an enemy attacks or undermines your product, this enemy usually spread lies since your product will negatively impact your enemy. But when the creator of a product is afraid of it, you should be worried. And unfortunately, this is the case here.

Humans were always eager on development and innovation but suddenly a lot of people in the tech market are calling for a time out. This has never happened before in the technology industry. The only cases we could remember when a global call was issued for a research shut down or pause, it was always something with horrible impact on mankind. Does Nuclear Arms ring a bell? Unfortunately, we learnt horrors of Nuclear Arms after the Hiroshima and Nagasaki Nuclear bombing. We are not ready to try something even more dangerous than Nuclear Arms if what Elon Musk thinks became true?

?After the introduction of the Internet in 1990s and further on the Social Media in 2000s among with the digitalization of great part of human civilization, we finally had enough data. This was Big Data breakthrough which was achieved several years ago. Coupled with the development of capable hardware, educated manpower and mega finance, we finally reached a breakthrough in Artificial Intelligence. Humans have successfully created a generative AI.

Generative AI is a type of artificial intelligence (AI) system capable of generating text, images, or other media in response to commands. The most famous example of generative AIs is ChatGPT, Bard and Midjourney. For example, in ChatGPT, once you order it to make an essay on something, it is capable of generating an authentic essay. In Russia, a student managed to pass his PHD research using ChatGPT. Midjourney can create picture as per your text feedback. For example, in Midjourney, someone asked for pasta harvest.

So now I can teach my kids that we grow pasta? The level of disinformation in AI is scary. This is but a glimpse of the risks of misuse of AI and not the dangers from AI itself. We are going to discuss these subjects in the following editions of the AI Revolution Series. What we are more concerned about AI itself.

Artificial Intelligence

is the simulation of human intelligence in machines in order to perform tasks that require knowledge of concept or a process, hopefully making a machine think; perceiving, synthesizing, and inferring information. AI Scientists took this step by step. They first started with creating an AI with narrow mindset aimed specifically toward one application which is considered the first age of AI. There are three stages of Artificial Intelligence;

Artificial Narrow Intelligence is limited AI aimed toward a specific complex application that can replicate and further on surpass human performance in the same field. The best example of ANI is Deep Blue developed by IBM. Deep Blue was a chess player expert system that runs on a unique dedicated supercomputer that managed to beat world chess champion, Garry Kasparov. This AI machine surpassed any human in playing chess by guiding it to the best move of all possible moves thanks to its inhuman calculations, memory and fast processing. Humans have achieved this stage.

Artificial General Intelligence is deep AI that mimics human intelligence, thus, can accomplish any intellectual task a human can as it is the representation of generalized human cognitive abilities in software so that, faced with an unfamiliar task, the AGI system could find a solution. AGI can learn alone new inputs by simulating similar historical exposure. ?We have yet to achieve AGI but Microsoft issued an article on April, 2023 stating that ChatGPT 4 is showing sparks of Artificial General Intelligence.

Artificial Super Intelligence is a form of AI that exceeds human intelligence in every aspect due to its greater memory, faster data processing, analysis and decision making capabilities by manifesting cognitive skills and innovating thinking patterns by itself. AGI will develop into ASI but experts fear that AGI transform into ASI without human interference or knowledge. It would be the pinnacle of AI research. Since we will be dealing with a machine that exceeds its creator, we do not know what actions it will take.

While AI stages focus on AI performance in tasks, it has little description on AI exposure and response to environment. This is where we get to AI types:

We have successfully achieved Limited Memory in several AI using deep learning methodologies. But, surprisingly in 2023, Chat GPT 4 has achieved theory of mind of 9 years infant. Also, Lamda Google AI passed the Turing Test which is the test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human.

The AI experts has focused on two main aspects to enhance AI, first, the AI perceiving the environment and second, AI generating simulated results. For perceiving the environment, AI experts created AI aimed for each of human senses:

Now that AI is capable of perceiving the outer environment alone based on one sense alone for each AI. AI started to realize complex concepts. Most of these AI models are in the introduction phase but ChatGPT 4 has achieved infant Theory of mind and has shown a glimpse of Artificial General Intelligence. Few years and each of these AI will develop into a greater level. Actually now, with mainly text based input, the new breakthrough of Generative AI has started. Generative AI is an advanced Artificial Intelligence that can create an output specifically tailored based on user prompt. It creates a neural network through consumed learnt data in order to simulate and generate new and unique outputs. Generative AI is capable of generative diversified out based on the machine. It is not only Chat GPT & Midjourney as some people think. You will be shocked at the number of existing Generative AI:

There are more than one AI for each function, we only showed a singular AI for most to ensure that each of these are covered. Generative AI is capable of performing more than these basic actions. Chat GPT 4 cover a lot of professional aspects which we will discover in the next edition of this series.

So, here is the big question. What if we merge all the perceiving AI & Generative AI into a singularity? We will reach Artificial Super Intelligence. And if ASI is reach and controlled, you have a super power with unrealized capabilities far more dangerous a million soldier. That is why Generative AI is the first step towards the making of AI Super Power. On global stage, the West and East are leading the scary technological race that is not realized between the common. In 2017, President Vladimir Putin said that the nation that leads in AI will be the ruler of the world. Later on, in March, 2023, Russia and China signed a "Joint Statement between the People's Republic of China and the Russian Federation on Deepening the Comprehensive Strategic Partnership of Coordination in the New Era.

The threat from AI on humanity is also real. The problem is some tech experts cannot identify if an AI is lying or not. The concept of hiding information from its master is scaring a lot of people. Geoffrey Hinton lately said“If I have 1,000 digital agents who are all exact clones with identical weights, whenever one agent learns how to do something, all of them immediately know it because they share weights,” “Biological agents cannot do this. So collections of identical digital agents can acquire hugely more knowledge than any individual biological agent. That is why GPT-4 knows hugely more than any one person.” Geoffrey Hinton may have backed away from AI further development because he believes that he miscalculated or cannot perceive what if a single mind learn from millions of experiences. It is unprecedented in history and psychology cannot forecast it.

Stay put for the next edition of AI Revolution series !

References:

Microsoft research - https://arxiv.org/pdf/2303.12712.pdf

#patterns #egypt #tech #technology #ai #artificialintelligence #generativeai #ANI #AGI #artificialgeneralintelligence #artificialsuperintelligence #asi #geoffreyhinton #elonmusk #openai #google #chatgpt #chatgpt4 #bard #midjourney #dalle2 #jarvis #github #githubcopilot #gastrograph #lamda #theoryofmind #ToM #limitedmemory #reactivememory #samaltman #chatbot #synthesia #jukebox #coding #microsoft #airisks #arybale #neutouch #techwar #techcommunity #techindustry #techeducation #techdocs #usa #us #china #baidu #erniebot #russia #technologytrends #godfather