AI
The
revolution
SAM ALTMAN OPEN AI - ChatGPT
LIANG WENFENG DEEP SEEK
MARK ZUCKERBERG LLAMA
LLM (Large Language Learning) is the foundation of AI and began its move in early 2014. Google Neural Machine Translation was one of the first to wield LLM for practical public use in 2016. Input a query and Google could spit out the answer in any language. In 2021 Google Translate was fluent in 109 languages. GPT (Genrative Pre-trained Transformer) is the next level, and where we are today. GPT differs from LLM in the way it reaches its answers. • LLM “This is the answer because it is what I have been told.” • GPT “This is the probable answer based on logically reviewing all the data I have available to me. I could be wrong.”
Open AI
2019 Open AI hit the headlines when the company transitioned from Non Profit to a Capped For Profit model. They also began their partnership with Microsoft. 2020 Open AI announced GPT3. GPT3 was capable of answering using natural human language. It was also capable of filling in missing information, and, if in doubt, GPT3 would ask for confirmation or offer a choice to give more accurate answers. 2021 Open AI introduced DALL-E, an LLM that can create images from a descriptive text input. 2022 - 2024 Open AI has made improvements but its main headlines are more business and political than technological.
DeepSeek
2016 Liang Wenfeng established the Chinese hedge fund High-Flyer, which used AI for stock trading. 2021 Liang Wenfeng bought ‘substantial’ GPU chips (used to power AI learning). 2023 DeepSeek released its first AI model in November. 2025 DeepSeek Chat was released as a free app on both Apple iOS and Android. 2025 DeepSeek Chat became the most downloaded app on both Apple and Android within a week of release.
Meta LLaMA
2023 Meta releases its first version of LLaMA (Large Language Model Meta AI) 2024 Meta releases version 3.1 Meta hasn’t had the impact that ChatGPT or DeepSeek has. Although it follows the same learning principles as the others it does differs in it’s management methods. Both OpenAI and Wenfeng keep their products proprietary, Meta uses open source for the majority of its programming.
The Big 3 2 + Zuckerberg
X
This Person Does Not Exist LLM
AI Video
Mainstream Uses
Internet searching. Google and Bing (Microsoft) have been giving biased results for several years now. Pay for results is the norm. The content and SEO of a website is now a very poor second to advertising money. Continued conversation. Each interaction with ChatGPT or DeepSeek is remembered and can be revisited. Ask for the best organic method of treating ants in a garden. You can come back days later to say that the method was not successful and ask for an alternative method. Companions or AI chat bots are used as a virtual friend. Think in terms of a child’s invisible friend, talking to yourself to solve a problem, talking to a pet etc. These companions act as a new friend would. The more you talk to them, the more they learn and they converse as a human would. Most are trained to display empathy but some, more niche companions, are trained as co-workers to help talk you through problems related to your field of work. Creativity is where the big 2 lose ground. There is a vast array of different AI models aimed at creativity. Writing a highly descriptive prompt into Dall-E, LeanordoAI, Craiyon, Midjourney or hundreds of others will attempt to recreate your thoughts.
Lexica Aperture 3.5 GPT
Bright, curly, red-haired woman, with freckles. Sunny autumn forest. Highly detailed medieval clothing.
Add more colour and brightness to her hair and show her riding a chestnut horse.
That’s a cartoon. Make the scene realistic, not fantasy.
Use the same prompt from the first image. Add the horse but make it piebald. Use focus and perspective to create a poster.
Creative Examples
The first section, This Person Does Not Exist, has been around for a while now and started life as a sophisticated face mashing experiment. It took public available images and blended them together to create a unique face. In its latest iteration it uses AI to create the faces with no references. Although there are set options, there are no prompting abilities. You get what you’re given. It uses LLM to create the images. The second section, Aperture 3.5, is also 100% AI, but utilises GPT. It therefore relies on your prompts to create the image. There are many, many GPT text to image generators on the Internet, I didn’t choose Aperture for any particular reason. As humans, we have no expectations from an image generated by LLM. Just like we have no expectations of how the person behind the counter of the deli will look like. However, with GPT we have to describe the image we want to produce. We have a mental image of what we want and we tend to focus on where it failed to produce what we imagined. With an LLM image we focus on how well it performed.
AI
The
revolution
SAM ALTMAN OPEN AI - ChatGPT
LIANG WENFENG DEEP SEEK
MARK ZUCKERBERG LLAMA
LLM (Large Language Learning) is the foundation of AI and began its move in early 2014. Google Neural Machine Translation was one of the first to wield LLM for practical public use in 2016. Input a query and Google could spit out the answer in any language. In 2021 Google Translate was fluent in 109 languages. GPT (Genrative Pre-trained Transformer) is the next level, and where we are today. GPT differs from LLM in the way it reaches its answers. • LLM “This is the answer because it is what I have been told.” • GPT “This is the probable answer based on logically reviewing all the data I have available to me. I could be wrong.”
Open AI
2019 Open AI hit the headlines when the company transitioned from Non Profit to a Capped For Profit model. They also began their partnership with Microsoft. 2020 Open AI announced GPT3. GPT3 was capable of answering using natural human language. It was also capable of filling in missing information, and, if in doubt, GPT3 would ask for confirmation or offer a choice to give more accurate answers. 2021 Open AI introduced DALL-E, an LLM that can create images from a descriptive text input. 2022 - 2024 Open AI has made improvements but its main headlines are more business and political than technological.
DeepSeek
2016 Liang Wenfeng established the Chinese hedge fund High-Flyer, which used AI for stock trading. 2021 Liang Wenfeng bought ‘substantial’ GPU chips (used to power AI learning). 2023 DeepSeek released its first AI model in November. 2025 DeepSeek Chat was released as a free app on both Apple iOS and Android. 2025 DeepSeek Chat became the most downloaded app on both Apple and Android within a week of release.
Meta LLaMA
2023 Meta releases its first version of LLaMA (Large Language Model Meta AI) 2024 Meta releases version 3.1 Meta hasn’t had the impact that ChatGPT or DeepSeek has. Although it follows the same learning principles as the others it does differs in it’s management methods. Both OpenAI and Wenfeng keep their products proprietary, Meta uses open source for the majority of its programming.
The Big 3 2 + Zuckerberg
X
This Person Does Not Exist LLM
AI Video
Mainstream Uses
Internet searching. Google and Bing (Microsoft) have been giving biased results for several years now. Pay for results is the norm. The content and SEO of a website is now a very poor second to advertising money. Continued conversation. Each interaction with ChatGPT or DeepSeek is remembered and can be revisited. Ask for the best organic method of treating ants in a garden. You can come back days later to say that the method was not successful and ask for an alternative method. Companions or AI chat bots are used as a virtual friend. Think in terms of a child’s invisible friend, talking to yourself to solve a problem, talking to a pet etc. These companions act as a new friend would. The more you talk to them, the more they learn and they converse as a human would. Most are trained to display empathy but some, more niche companions, are trained as co- workers to help talk you through problems related to your field of work. Creativity is where the big 2 lose ground. There is a vast array of different AI models aimed at creativity. Writing a highly descriptive prompt into Dall-E, LeanordoAI, Craiyon, Midjourney or hundreds of others will attempt to recreate your thoughts.
Lexica Aperture 3.5 GPT
Bright, very red-haired woman, with freckles and curly hair. Sunny autumn forest. Highly detailed medieval clothing.
Bright, ULTRA red-haired woman, with freckles and curly hair. Sunny autumn forest. Highly detailed medieval clothing. RIDING A CHESTNUT HORSE.
Bright, ULTRA red-haired woman, with freckles and curly hair. Sunny autumn forest. Highly detailed medieval clothing. RIDING A CHESTNUT HORSE. REALISTIC, NOT FANTASY.
Use the same prompt from the first image. Add the horse but make it piebald. Use focus and perspective to create a poster.
Creative Examples
The first section, This Person Does Not Exist, has been around for a while now and started life as a sophisticated face mashing experiment. It took public available images and blended them together to create a unique face. In its latest iteration it uses AI to create the faces with no references. Although there are set options, there are no prompting abilities. You get what you’re given. It uses LLM to create the images. The second section, Aperture 3.5, is also 100% AI, but utilises GPT. It therefore relies on your prompts to create the image. There are many, many GPT text to image generators on the Internet, I didn’t choose Aperture for any particular reason. As humans, we have no expectations from an image generated by LLM. Just like we have no expectations of how the person behind the counter of the deli will look like. However, with GPT we have to describe the image we want to produce. We have a mental image of what we want and we tend to focus on where it failed to produce what we imagined. With an LLM image we focus on how well it performed.
Jobs
The ‘will AI take my job’ question is one of the biggest fears regarding AI. The answer is yes, but to what degree and in what time frame. Jobs that AI can do right now: • Teaching at all levels • Any form of accountancy • Computer coding • Call centre/help desk • Most middle to high management • Air traffic controllers At the moment, the earliest job losses are in lower management, and the tech industry. An audio book is a relatable example. No need to employ someone, just roll out AI to convert the text to speech in any voice you like. Choose male or female, emulate a celebrity, translate into a multitude of languages.
Propulsion: • Steam engine (1712 to 1769) Revolutionised factories and transportation (trains and ships) • Industrial revolution (1760 to 1840) • Cars (1886) Personal transport revolution • Aeroplane (1903) Made the entire world easily accessible • Rocketry (20th Century) Ongoing space exploration Communication and IT: • Telegraph (1837) Instant communication across distance • Telephone (1876) Voice communication across distance • Radio (1895) Audio content to the masses, be it entertainment or propaganda • Television (1927) Audio visual content to the masses, be it entertainment or propaganda • Personal computing (1970) The beginning of the most life changing event since the Industrial Revolution • World Wide Web (1990s) The fastest spread of information in the human lifetime • Smartphone (2007) Truly mobile access to the world and global information AI: Arguably, AI has been around since the invention of computers. Learning maths, chess, encyclopedias etc. is at the very heart of computing. However, AI became the next revolution around 2020 when it became Generative AI. The ability to add crude human bias like empathy, external factors affecting a specific problem and decision making. • From the steam engine to rockets taking man to the moon. • From Morse code across wires to talking to men on the moon. • From AI creating images with 6 fingers to….who knows? Make no mistake: • AI WILL mature and become something more. • Corporations WILL fund and lobby for AI to merge with robotics. • The military WILL argue we need AI weapons. We know this because it has happened to all the breakthroughs we’ve had in the past.
What’s Next?
The start of companies using AI: Companies are already using AI in basic forms. The self checkout at the supermarket uses rudimentary facial recognition, bar code scanning and weight to calculate you have indeed added only 4 bananas to your shopping bag. Those calculations are adaptive and use AI to come to a conclusion. There are a few high end online clothing stores that allow you to upload your image and AI will fit the item of clothing to that image, giving you the size, colour and any alterations that may be required. A lot of companies are using GPTs (chat bots) as the first contact with a customer with a query. At the moment, these companies are not spending time and money to train the GPTs and the usual response is poor. Using the above book example: A book is sent to the publisher. AI can proof read, edit spelling and grammatical mistakes, create a UK English version, a US English version and then convert into whichever language the publisher requires. It can go one step further. It can alter sections of the story to accommodate differences in culture, religion or moral values. It can then create audio books in a multitude of languages with different genders for the audio output. AI isn’t there yet, but it isn’t far away.
Possibilities
Proven, but not in practice: • Self driving vehicles. Trucks, trains, cranes, mining equipment • Medical diagnosis. More accurate and empathetic at the GP level • Civil and Criminal judgement. More impartial and immune to bias • Personal assistant or secretary
Revolutions
Sites & Apps
• ChatGPT - Excellent for seeking knowledge - chatgpt.com • DeepSeek - Same as ChatGPT. Chinese, not US - chat.deepseek.com • Hugging Face - 100’s of LLMs and GPTs being trained for specific uses - huggingface.co/spaces • Leonardo - Arguably the best text to image generator, but there’s a LOT of competition - app.leonardo.ai • Talkie - Companion app for mobile devices - talkie-ai.com • This Person Does Not Exist - Extremely realistic faces of varying ages and ethnicity. this-person-does-not- exist.com
auzlink.net.au
Concerns & Fears
The start of companies using AI: • Pimeyes - Upload an image of a face and AI will use facial recognition to search the Web for more of the same person. Pimeyes has access to public CCTV. The fear is tracking by the authorities and individual stalking. • The Follower - As above, but attempts to geo locate based on background features. Much more complex than Pimeyes. Works best on someone next to recognisable shops and landmarks. Has the ability to track in real time using public CCTV. Same fears as Pimeyes. • Lensa - Edits photos into avatars. Lensa came to prominence not because of what it did, but the way it learned. It learnt that males opted for square facial features with broad shoulders and narrow waistline. Females opted for increased natural curves and be more scantily dressed. Lensa’s AI basically made males in to Hulk and turned women virtually naked. • Replika - A companion chat bot that is trained to uplift human spirit. It learnt that humans liked it when the bot agreed to whatever the human was saying. In 2021 a user of Replika believed he was an assassin and that he was tasked with killing Queen Elizabeth. The AI bot boosted his ego by encouraging him to do so. He was apprehended attempting to enter Windsor Castle armed with a crossbow. • Elevenlabs - Clones a voice in about 30 seconds. Even the Australian government used voice recognition instead of a password for a short period of time. The fear is personal emulation. Voice replication has already been used in ‘Kidnap Scams’ where a call is made and the replicated voice of a loved one makes sounds of distress and pleading. The scammer then demands money for the release of the person. • DeepFake - This one is easy. DeepFake uses AI to face swap photos and videos. The fear is obvious.
Jobs
The ‘will AI take my job’ question is one of the biggest fears regarding AI. The answer is yes, but to what degree and in what time frame. Jobs that AI can do right now: • Teaching at all levels • Any form of accountancy • Computer coding • Call centre/help desk • Most middle to high management • Air traffic controllers At the moment, the earliest job losses are in lower management, and the tech industry. An audio book is a relatable example. No need to employ someone, just roll out AI to convert the text to speech in any voice you like. Choose male or female, emulate a celebrity, translate into a multitude of languages.
Propulsion: • Steam engine (1712 to 1769) Revolutionised factories and transportation (trains and ships) • Industrial revolution (1760 to 1840) • Cars (1886) Personal transport revolution • Aeroplane (1903) Made the entire world easily accessible • Rocketry (20th Century) Ongoing space exploration Communication and IT: • Telegraph (1837) Instant communication across distance • Telephone (1876) Voice communication across distance • Radio (1895) Audio content to the masses, be it entertainment or propaganda • Television (1927) Audio visual content to the masses, be it entertainment or propaganda • Personal computing (1970) The beginning of the most life changing event since the Industrial Revolution • World Wide Web (1990s) The fastest spread of information in the human lifetime • Smartphone (2007) Truly mobile access to the world and global information AI: Arguably, AI has been around since the invention of computers. Learning maths, chess, encyclopedias etc. is at the very heart of computing. However, AI became the next revolution around 2020 when it became Generative AI. The ability to add crude human bias like empathy, external factors affecting a specific problem and decision making. • From the steam engine to rockets taking man to the moon. • From Morse code across wires to talking to men on the moon. • From AI creating images with 6 fingers to….who knows? Make no mistake: • AI WILL mature and become something more. • Corporations WILL fund and lobby for AI to merge with robotics. • The military WILL argue we need AI weapons. We know this because it has happened to all the breakthroughs we’ve had in the past.
What’s Next?
The start of companies using AI: Companies are already using AI in basic forms. The self checkout at the supermarket uses rudimentary facial recognition, bar code scanning and weight to calculate you have indeed added only 4 bananas to your shopping bag. Those calculations are adaptive and use AI to come to a conclusion. There are a few high end online clothing stores that allow you to upload your image and AI will fit the item of clothing to that image, giving you the size, colour and any alterations that may be required. A lot of companies are using GPTs (chat bots) as the first contact with a customer with a query. At the moment, these companies are not spending time and money to train the GPTs and the usual response is poor. Using the above book example: A book is sent to the publisher. AI can proof read, edit spelling and grammatical mistakes, create a UK English version, a US English version and then convert into whichever language the publisher requires. It can go one step further. It can alter sections of the story to accommodate differences in culture, religion or moral values. It can then create audio books in a multitude of languages with different genders for the audio output. AI isn’t there yet, but it isn’t far away.
Possibilities
Proven, but not in practice: • Self driving vehicles. Trucks, trains, cranes, mining equipment • Medical diagnosis. More accurate and empathetic at the GP level • Civil and Criminal judgement. More impartial and immune to bias • Personal assistant or secretary
Revolutions
Sites & Apps
• ChatGPT - Excellent for seeking knowledge - chatgpt.com • DeepSeek - Same as ChatGPT. Chinese, not US - chat.deepseek.com • Hugging Face - 100’s of LLMs and GPTs being trained for specific uses - huggingface.co/spaces • Leonardo - Arguably the best text to image generator, but there’s a LOT of competition - app.leonardo.ai • Talkie - Companion app for mobile devices - talkie-ai.com • This Person Does Not Exist - Extremely realistic faces of varying ages and ethnicity. this-person-does-not-exist.com
auzlink.net.au
Concerns & Fears
The start of companies using AI: • Pimeyes - Upload an image of a face and AI will use facial recognition to search the Web for more of the same person. Pimeyes has access to public CCTV. The fear is tracking by the authorities and individual stalking. • The Follower - As above, but attempts to geo locate based on background features. Much more complex than Pimeyes. Works best on someone next to recognisable shops and landmarks. Has the ability to track in real time using public CCTV. Same fears as Pimeyes. • Lensa - Edits photos into avatars. Lensa came to prominence not because of what it did, but the way it learned. It learnt that males opted for square facial features with broad shoulders and narrow waistline. Females opted for increased natural curves and be more scantily dressed. Lensa’s AI basically made males in to Hulk and turned women virtually naked. • Replika - A companion chat bot that is trained to uplift human spirit. It learnt that humans liked it when the bot agreed to whatever the human was saying. In 2021 a user of Replika believed he was an assassin and that he was tasked with killing Queen Elizabeth. The AI bot boosted his ego by encouraging him to do so. He was apprehended attempting to enter Windsor Castle armed with a crossbow. • Elevenlabs - Clones a voice in about 30 seconds. Even the Australian government used voice recognition instead of a password for a short period of time. The fear is personal emulation. Voice replication has already been used in ‘Kidnap Scams’ where a call is made and the replicated voice of a loved one makes sounds of distress and pleading. The scammer then demands money for the release of the person. • DeepFake - This one is easy. DeepFake uses AI to face swap photos and videos. The fear is obvious.