Artificial Intelligence, a topic that's really hot nowadays. People can't stop talking about it, especially after OpenAI launched ChatGPT3 in November 2022. However, the technology itself is not new, it has only recently gained mainstream attention. Many people have started questioning how safe AI would be and how it will shape our future. Will it replace our jobs? Is it going to be evil?
In this post, I want to take a moment to think about this new technology, share my personal experiences, and discuss the way I would like to see it progress in the future. Let's embrace these new technologies, work together to find solutions, and remain optimistic about our future to empower ourselves.
AI vs Generative AI
AI, or artificial intelligence, is a term used to describe computer systems that can perform tasks typically requiring human intelligence. Recently, a new type of AI has gained popularity known as GAI, or generative artificial intelligence. While traditional AI is designed for repetitive and specific tasks such as image recognition, self-driving cars, and voice assistants, Generative AI has the ability to learn patterns from training data and generate new data that has never been seen before. This includes text, images, audio, and other forms of data. To illustrate, consider this entertaining example of the pope in a big puffy jacket:
The Rise of AI
The AI hype hit its peak in late November 2022 when OpenAI unveiled GPT3. After just 5 days the app hit over 1 million users. Now, why did the technology explode so much if I tell you that AI already existed in the 90’s?
The answer lies in NLP (natural language processing), a subpart of AI that enables computers to understand spoken or written language. ChatGPT is an LLM (large language model) which is a subpart of NLP where it can interpret the input given and respond by generating appropriate new answers. This gives us the availability to have conversations in which we interact with the computer. Just like having a conversation with any another person. By this, OpenAI gave us an easy help app in which we can ask for help just as we could do to another real person. The big difference is that ChatGPT knows a lot more than even the smartest human in the world.
Another key factor fueling AI's growth is the advancement in computer chips. The latest breakthroughs provide computers with the necessary computation power to handle complex AI tasks. For example, ChatGPT was trained on 570GB of datasets, including web pages, books and other sources. They used 1023 GPU’s (a type of computer chips) and the training process was completed in approximately 34 days.
Building AI is a long process
To build this kind of technology you require a lot of time and money. Let’s see some numbers about the creation of OpenAI GPT3.
Some fun numbers to know of GPT
- GPT3 has 175 billion parameters and receives around 10 million queries a day
- Daily cost of running ChatGPT to be approximately $100,000, totaling $3 million per month.
- The cost of training GPT3 was around 5 million and took roughly 34 days to complete.
- Sam Altman (CEO OpenAI) revealed in an interview that the cost of training GPT4 is more than 100 million.
Source: https://medium.com/codex/gpt-4-will-be-500x-smaller-than-people-think-here-is-why-3556816f8ff2
Applications utilizing GPT3
- DALL-E: creating images from text.
- CLIP: connecting text and images.
- Whisper: multi-lingual voice-to-text.
- ChatGPT: chatbot, article writer, code writer.
If you’re interested in the journey and more technical side of ChatGPT, I recommend this article: https://medium.com/walmartglobaltech/the-journey-of-open-ai-gpt-models-32d95b7b7fb2
Other alternatives of ChatGPT are: Claude 2 (only in US and UK), Bard
AI Agents
Auto-GPT represents an "AI agent" that receives a goal in natural language by a real human and will attempt to achieve it by breaking it into sub-tasks and using the internet and other tools in an automatic loop. It uses OpenAI's GPT-4 or GPT-3.5 APIs, and is among the first examples of an application using GPT-4 to perform autonomous tasks. Unlike interactive systems such as ChatGPT, which require manual commands for every task, Auto-GPT assigns itself new objectives to work on with the aim of reaching a greater goal, without a mandatory need for human input. It is able to execute responses to prompts to accomplish a goal task, and in doing so will create and revise its own prompts to recursive instances in response to new information. Auto-GPT manages memory using databases and files, addresses input length restrictions using summarization, performs internet-based actions like web searches and form filling on its own, and can produce voice output through text-to-speech capabilities.
Example: Ordering Pizza
One of the tasks that AutoGPT can help you with is ordering a pizza. To order a pizza with AutoGPT, you need to define your role and goals. For example, your role could be “pizza lover,” and your goals could be “find the best pizza place near me, ”order a large pepperoni pizza with extra cheese,” and “pay with a credit card.” AutoGPT will then use its internet access, memory management, and text generation capabilities to perform the following steps:
- Search for pizza places near your location and compare their ratings, prices, and reviews.
- Select the best pizza place based on your preferences and availability.
- Call or visit the website of the pizza place and place your order using natural language.
- Confirm your order details, delivery address, and payment method.
- Wait for your pizza to arrive, and enjoy!
So all this without human interaction!
Another live example video:
so @united lost my luggage again
— harpriya (@harpriiya) July 20, 2023
I used an AI agent to call them with my info and track my bag down lol pic.twitter.com/iIbdY4ZwUd
You and AI as a team
As AI continues to grow and impact our lives, it won’t replace humans. AI will solve larger problems at a much faster rate than humas can do. Only a couple of things I’m not pretty sure of is creativity, intuition and critical thinking. These skills are really hard and require a lot of practice to completely master it.
When AI and humans team up, they create an unstoppable team. In healthcare, AI helps doctors make better diagnoses by analyzing medical data. This means we get better treatment and care. In business, AI can analyze market data, giving us valuable insights to make smarter decisions and grow our companies. One cool thing about this partnership is how it powers creativity. AI can handle boring, repetitive tasks, giving us more time to be creative. It even generates art and music, inspiring human artists to create amazing pieces based on AI's ideas.
While all of this looks promising, we need to be careful too. Fairness, transparency, and privacy protection are vital in making AI systems. We must address any biases in AI algorithms and have human oversight to choose the final outcomes.
As AI becomes a bigger part of our lives, we need to keep learning and adapting. With the help of AI we can adapt more knowledge quicker. A good example that I personally use is letting AI make quizzes. When I learn a new topic I always let ChatGPT make a quiz about it so that I’m sure I fully understand the concept. By staying curious and open to new technologies, we can stay ahead and make the most of this AI-human partnership.
Will we all die?
Probably not, however AI can still be very dangerous and it’s crucial to make proactive initiatives to reduce it’s bad behavior. We can to this by:
-
Trust & Ethics : We need to be transparent about how we train AI models and how it works. This result in having trust in each other.
-
Regulations: global guidelines for AI usage is a real challenge, given the diversity of countries and their laws. However, establishing fundamental red lines and regulations worldwide is essential to govern AI's ethical application.
A good example of responsible AI implementation is OpenAI's decision not to open source (making code public) ChatGPT's code. The reason behind this is to prevent potential misuse by developers with bad intents. An example of this was ChaosGPT. An AI tool that’s seeks to destroy humanity.
To be honest, almost all technologies started as something crazy and dangerous for some people. When the internet came, a lot of sceptics and criticism was going through the world. It would also been bad for humanity and destroy us. And look where we are now, we live in a society where it is crazy if you don’t have internet access.
Conclusion
AI and GAI already play a big part in our lives, even if you think you never used it before. It’s hardly visible but yet so powerful. Rather than fear or await its developments, we must embrace, learn more, and take AI to a direction where we feel comfortable using it. So what will the future bring us? The technology will make a shift in how we work/behave just like the internet did or smartphones. AI will make our lives more easy and comfortable but not lazy.