Though this is an honest review, we may earn a commission if you buy this product through our link. (learn more).
Scientists are working on a way to let AI programs develop their own algorithms. This would be a huge step forward.
Science fiction has predicted credit cards, TV and the 1969 lunar landing. Bionic limbs, military tanks and antidepressants have also emerged from the genre.
Some AI image generators allow users to create realistic art using text prompts. Others expand pictures beyond their original borders and use a variety of different styles.
Has AI Created Another AI Yet?
AI is making major strides in business and everyday life. Companies spend billions on AI products and services, universities make it a part of their curricula, and the Department of Defense is upping its game. AI is having a massive impact on global business operations, cybersecurity, medical diagnostics, and countless other fields. It also offers benefits such as reduced costs, faster time to market and more.
But one big question that has been posed is whether AI can actually create other AIs. Yes, there have been instances of AI creating other AIs or “Child AIs” but these AIs are created with a specific goal in mind by their creators.
For example, an AI called AutoML developed by Google was designed to help other AIs develop machine learning algorithms. The AI itself isn’t able to understand complex concepts like physics or math but it can learn to identify patterns in data and improve its own code. The result is a machine learning algorithm that can perform the task of developing other machine learning algorithms more effectively than humans could.
While these AIs are great successes, they still don’t come close to matching or eclipsing human intelligence. It’s unlikely that the scientific community will ever fully understand how the human brain works to the point where it can achieve true artificial general intelligence.
One field of AI research that holds promise for achieving artificial general intelligence is machine learning. It’s an area of AI that has been around for decades but is now being used to teach machines how to learn and adapt. ML uses reinforcement learning to identify patterns and improve its own performance. It’s a process that requires vast amounts of data and can be tedious but it has shown promise in areas such as machine translation, face recognition, and autonomous vehicles.
Researchers are now combining ML with another technique called deep neural networks to create AI that can recognize objects in images and predict what those objects will do next. This type of AI is called generative adversarial networks or GANs and it’s a powerful tool that has the potential to solve a wide variety of problems such as fraud detection and computer vision.
Is This Science-Fiction or Reality?
There have been many attempts to create artificial intelligence over the centuries. The Greek gods were depicted forging robot-like servants, engineers in ancient Egypt built statues of gods animated by priests, and thinkers like Aristotle, Ramon Llull, and Rene Descartes described the process of thinking as a pattern of cause and effect. Modern computer programmers have been using those concepts to create AI systems. They have used machine learning to develop algorithms that can perform tasks humans do, including identifying patterns, interpreting speech, playing games, and solving complex problems. They have also incorporated logic programming languages and other AI tools to build systems that can learn without supervision.
The current decade has seen the development of generative AI, which can produce new content by processing inputs such as text, images, designs, videos or musical notes. These systems typically begin with a prompt, such as an essay topic or an answer to a question, and then return the output it believes best matches the input. Generative AI is capable of producing a wide variety of products, from sophisticated fakes that can pass the Turing test to detailed medical diagnoses and financial investment recommendations.
Despite this progress, it is still difficult to create AI that is truly intelligent. This is because the algorithms that make up most AI are based on patterns and observations rather than logical reasoning, which means they tend to rely heavily on experience and biases to guide their decisions. It is important for computer programmers to carefully examine the decisions of their AI systems to ensure they are fair and non-discriminatory. Failure to do so could lead to algorithms that are unable to solve certain problems or create harmful outcomes.
Some AI is already part of our everyday lives, such as voice-activated smart speakers that can help us with daily tasks and research. AI is also being used in healthcare, such as a software tool called Copilot that helps doctors perform radiology procedures more quickly and efficiently. It is even transforming the automobile industry, with autonomous cars being developed by companies like Tesla, Apple and Google that will soon be on our roads. AI is also improving products we use at home and work, such as smart cameras, automation software and robotic vacuum cleaners.
Could It Soon Happen?
As artificial intelligence (AI) continues to improve, it can be used in a variety of ways. From personal assistants that manage calendars and reminders to software development tools that speed up the process of writing code, AI is improving people’s lives and making their jobs easier.
The technology has been embraced by many industries, especially in business where it is commonly used to automate repetitive tasks like customer service or fraud detection. It also helps to reduce human error and increase productivity. However, some experts worry that AI could eventually replace humans in some jobs.
While the fear of an AI takeover is real, it’s unlikely to happen soon, largely due to the massive computational and technical data infrastructure needed to run the systems. Nevertheless, AI technology can create a number of new jobs by helping to automate manual processes and freeing up human workers to perform more complex duties.
One of the most popular applications of AI is in chatbots that use large language models to provide human-like responses to questions. These models are trained on a massive amount of text to learn how words and phrases interact to make up sentences. One of the most well-known is OpenAI’s ChatGPT, which has wowed users and is now part of Microsoft’s search engine, as well as Google’s competing chatbot, Bard.
Another area where AI is making significant gains is in autonomous systems. The ability to use machine learning to train robots to perform a task has led to some impressive examples, including Boston Dynamics’ robots that can move in a range of terrains and even dance.
There are also a number of companies offering AI as a service, which can streamline the data engineering and model development tasks that are necessary to weave this type of technology into existing apps or to build new ones from scratch. Among the most prominent are AWS AI Services, Google Cloud AI platform and IBM AI solutions.
The ultimate goal of many AI researchers is to create a system that can solve problems it hasn’t been trained to work on, much like a human brain. This is known as “strong AI” or artificial general intelligence. While we are a long way from the kind of intelligence seen in movies like Westworld or Star Trek: The Next Generation, pioneers like DeepMind’s AlphaGo are making significant strides toward that goal.
Are We Ready for It?
Despite the concerns of some, most experts remain optimistic about the potential for AI to revolutionize how we live. Many of them have focused on health care, predicting that AI will help identify and treat conditions such as cancer and Alzheimer’s disease, enabling people to lead longer, healthier lives. Other experts have hailed the ability of AI to improve education by analyzing big data and providing customized learning materials for each student’s unique needs.
The current decade has also seen the advent of generative AI, which can take a variety of inputs such as text, image or audio and produce new content based on that prompt. The resulting output can be anything from essays and solutions to problems to realistic fakes created from photos or video of people. These generative AI tools are becoming popular toys, but they can have serious business applications as well. For example, a company could use them to produce a variety of marketing or technical writing in seconds, then revise it quickly based on feedback.
Other researchers are working on using generative AI to develop better medical treatments and a range of other industries. For instance, a team at Stanford University has used the technology to create 3D models of human brains that can be used for surgery and research purposes. This is a key step toward building better prosthetics and other medical devices.
Some are even betting that generative AI can eventually help solve complex global problems such as climate change, war and poverty by identifying patterns and trends in historical data. They believe that by predicting the outcome of natural disasters, weather forecasts and food production, it will be possible to prepare for them and mitigate their effects with more effective early warning systems.
It may seem far-fetched that an AI can create another AI, but Wang and others like him believe it could be a shortcut to creating supersmart machines. After all, seven decades after the first serious efforts to build artificial intelligence began, we’re still a long way from creating an AI that’s as smart as humans.