AI amidst technological revolution

7 Decades later we have now reached the state of having advanced conversational AI models able to engage in a wide range of conversations with the ability to answer questions provide complicated explanations and assist in multiple tasks ranging from solving math equations to write programming code. AI today is used in almost any person with internet access to assist with simple tasks with chat-GPT being the most famous one.

But why now? Which parameters lined up perfectly to make AI possible to exist. The main limitation we had in the past was the computational performance of our electronics. To put it into perspective, from 1971 which marks the day of the first commercially produced microprocessor to today, processing power has increased roughly 100 million times for processors and 20 billion times if we compare it with GPU (Graphics Processing Units) widely recognized and commercialized in the late 90s.

Science is mostly driven by investments through R&D for companies that create new technologies and thus by analysing the stock prizes someone can have a decent insight which areas are developing or tend to develop in the following years. In the previous paragraph we mention GPUs that have immense processing capabilities using parallel computations making them the most efficient when it comes to neural networks training. Mainly for that reason, Nvidia having invested heavily on their CUDA architecture and advanced parallel processing capabilities while also maintaining 75% of the market share of GPU manufacturing became the most evaluated company worldwide in a matter of months with their stock rapidly rising.

AI has heavily infiltrated every aspect of science being used for research purposes. The Nobel prizes for Physics and Chemistry both were awarded this year to individuals for their contributions to artificial neural networks for the former and protein structure prediction and computational protein design for the latter. This stands as a testament to how highly Artificial intelligence is viewed by the scientific community. Which begs the question for the next paragraph of our article “How can AI being implemented in the future and what types of future technologies may arise from that?”.

We can start by presenting the AI possibilities in medicine and disease diagnostics. In every case it is always optimal to treat disease before it has progressed to a more advanced stage and that cannot be further from the truth when it comes to cancer. The survival rate of every cancer type significantly decreases as the disease progresses to advanced stages making the need of a blood probing very promising for battling this disease that kills millions every year while also burdening the national health system of every country. AI can be used to found patterns between the blood markers to identify the disease much earlier and that can also be the case for other diseases like diabetes, neurological disorders and cardiovascular diseases.

KAI can be used to found patterns between the blood markers to identify the disease much earlier and that can also be the case for other diseases like diabetes, neurological disorders and cardiovascular diseases. Image: Vasileios Papalampropoulos with Bing Image Creator.

On the context of cardiovascular diseases, AI can be implemented predict a heart attack before it even happens by using data collected by wearable technology. By 2024 roughly 30% to 40% of Americans are using smartwatches or fitness trackers containing health monitoring features able to measure the user’s heartbeat throughout the day. By using the data collected from smartwatches belonging to patients that have suffered a heart attack we might be able to create a model able to recognize a pattern to those individuals’ heartrates, then we can use the same model to identify and then warn a new user that a heart attack might be imminent in the following hours or even days before this happens incentivising the user to seek medical attention.

AI can be used for advance simulations in material science helping scientists create exotic materials with extreme mechanical properties. Those materials can be used in various applications like aerospace engineering as well as for civil engineering. Image: Vasileios Papalampropoulos ChatGPT / DALL-E.

Aside from the breakthroughs in the medical field, AI can be used for advance simulations in material science helping scientists create exotic materials with extreme mechanical properties. Those materials can be used in various applications like aerospace engineering as well as for civil engineering. The days where a high magnitude earthquake could level an entire city might be a thing of the past in the near future where more durable materials that are also economically viable may be used in a large scale. Solar panels able to harvest a large percentage of the solar radiation can help the transition to green energy by being implemented in house roofs to relief the burden in the network during the high demand present in the middle of the day. Renewable energy is completely dependant to physical phenomena of earth’s weather especially when it comes to solar and wind power and thus can be extremely unreliable for extensive use. A way to counter this is to connect multiple sources of renewable energy like solar, wind and hydroelectricity and then with the use of sophisticated microcontroller monitoring that capture the solar activity and the wind we can use an AI that will control the function of the hydroelectric dam. Hydroelectric power can quickly ramp up or down to meet demand when solar or wind output fluctuates thus having a reliable power network by releasing water from the reservoir when the solar, wind output is low and storing water when the output is high.

Self-driving cars have already shown us very promising results with the level of automation reaching high levels even today. When the models and the technology reach safety levels that might surpass the human drivers and that is something we will see throughout the 2 next decades, we might start to see a slow transitioning to self-driving cars caused by basic economical motives combined with some functionality benefits. Self-driving cars will become cheaper to ensure and will have the ability to be driven more fuel efficiently thus making them more attractive to the consumer based on. Another advantage is the ability for underage users to use the vehicle without having the need for a driver license thus making the responsibility to “take the kids to school” a thing of the past. The ability for the vehicle to be used for a night out without the concern of exceeding the amount of alcohol can also be considered a plus for many.

In the far future, when space travel might be more accessible, we can benefit from autonomous navigation making real time decisions during flight. Image: Vasileios Papalampropoulos with Bing Image Creator.

In the far future, when space travel might be more accessible, we can benefit from autonomous navigation making real time decisions during flight. Space missions can generate vast amounts of data requiring the use of AI to help scientists identify patterns, anomalies, and potential discoveries in fields like astrophysics and planetary science. In the farfetched scenario that we manage to achieve interstellar travel monitoring spacecraft systems to predict potential failures before they occur while also managing systems over long durations and distances without constant human oversight is necessary and can implemented with AI. More realistically during this century we can use artificial intelligence for an upcoming mission to mars to analyse the data collected by the crew sent there or alternatively to control robots able to navigate challenging terrains, conduct experiments, and make decisions without direct human intervention. AI interfaces inside astronaut suits can actively monitor their health, managing life support systems, and providing decision support in high-stress situations.

What about the future of computing using AI tools? Already existing technologies like classical computers can become increasingly faster and better with the introduction of composite materials that are better at handling temperatures and other scaling issues that silicon faces in modern transistors. But this can only go so far, we have already mostly pushed the semiconductor technology leaving behind a small room for improvement, we are not going to see 100-million-fold increase in computation power as we experienced in the last decades. So, if semiconductor technology isn’t going to be improved to the same extent we need to investigate new computation technologies like quantum computing. Quantum computing started as a concept around the early 70s roughly the same time microprocessor technology started. It was further developed in the 1980s when quantum computing started to catch the attention of famous physicists like Richard Feynman that gave an MIT talk discussing the inability of classical computers to simulate the evolution of a quantum system proposing a basic model for a quantum computer. By then efforts has been made to improve quantum error correction, developing more stable qubit technologies, and exploring applications in cryptography, material science, and complex problem solving. In many modern-day algorithms that require polynomial time such as Shor’s Algorithm a sufficiently powerful quantum computer can solve it in a matter of hours or days with a classical requiring millions of years. AI can assist in designing and understanding quantum algorithms and systems by modelling quantum phenomena while also expedite the discovery of new materials for quantum computing, such as better qubit designs or error correction methods.

Will it one day be possible to beam people from one place to another? Sounds difficult, I t cannot be ruled out tht AI will play a deciive role in this. Image: Vasileios Papalampropoulos mit ChatGPT / DALL-E.

One thing is sure, AI will completely change the way scientific progress is made completely revolutionizing research. No one knows what wait in the future and many of the technologies that were discussed here fall between implementation and speculation.