In today’s world, there is a lot of talk about artificial intelligence (AI). For years, Hollywood has explored the potential tragedy that AI could wreak, and billionaires like Elon Musk have spoken out about the benefits AI could bring to society. I was.
Products such as smartphones and their apps also incorporate some AI. Around the world, AI has become a term that seems to come out of nowhere, but is now just a part of everyday life.
But what is AI and how does it work? ) [AI**2]) said one thing needs to be clarified before answering these questions. It’s that the AI we see isn’t exactly AI.
“The current conversation revolves around deep learning technology, which is basically a very powerful learning model from big data. [sets]”
Artificial intelligence and deep learning
Deep learning technologies work by feeding computer software large amounts of labeled data, or data tagged to provide context about what it is. From there, the software can sift through the received data and make predictions on the unlabeled data.
Common examples of this include facial/fingerprint recognition software on smart devices, online chatbots like ChatGPT, and even Tesla self-driving cars with deep learning software.
However, the term AI has recently become an umbrella term that includes terms such as deep learning and machine learning technologies, algorithms, and more. According to IBM, artificial intelligence in practice refers to a type of computer software that “leverages computers and machines to mimic the problem-solving and decision-making abilities of the human mind.” These computers are built for the tasks they are asked to perform.
The current form of AI is very different from what is portrayed in films such as James Cameron’s. terminator from 1984. In this movie, computer software can adapt, learn, grow, and generally recreate the human mind and all its nuances.
Concerns about AI
The idea of this kind of AI can be seen over 50 years ago, when scientists were experimenting with neural networks that mimic the way the brain and its neurons send signals to each other. This idea of mimicking the human mind and its neural networks is still found in all technologies related to AI in one way or another.
These technologies are just the accumulation of older innovations that came before, but the datasets that feed them raise a new problem: online privacy.
Steve Drew, an assistant professor at the University of Calgary School of Software and Electrical Engineering, said traditional methods of collecting data for these technologies could also include personal information.
“like that [an] These companies with lots of data can of course do good things that benefit society as a whole, but they can also do bad things like surveillance of the public and predicting things people don’t want to predict. ‘ says Drew.
Google’s own privacy policy states that it collects information through the apps you operate, your phone number, and even the people you most frequently contact via Gmail.
Other businesses and web browsers may also ask for some personal information when you create a new account on their website, or financial information when you buy something online.
address problems and find solutions
Concerns about privacy and personal data collection have become such a big issue that governments are now enacting laws aimed at protecting user privacy and confidentiality, such as the European General Data Protection Regulation (GDPR). has passed.
But Drew’s research suggests that this problem can be solved using a different kind of AI learning software: federated learning, which allows software to receive data and train without collecting private data. It shows what you can do.
“Companies don’t necessarily need input history, but the idea is that they can learn from it without associating personal information with the user. This way privacy concerns are mitigated, but machine learning will continue. can.”
Other concerns surrounding the artificial intelligence industry in general include how collected data is retrieved, energy efficiency, and more. Souza believes the benefits outweigh the drawbacks in this case.
“I would be [more] how excited are you [AI is] It will impact healthcare delivery, robotics, and many other important areas that make the world a better place. “
Souza’s research, for example, is looking at ways to improve the speed at which MRIs are performed, which could also help patients get their scans sooner. By feeding the deep learning model a labeled dataset of her sick MRI scans and his healthy MRI scans, Souza says these scans can run faster overall, and detect problems faster. I think we can.
Artificial intelligence technology has its strengths and weaknesses, but its integration with our society will continue. Scientists and politicians are figuring out ways to make it safer across the board, but we’re far from worried that the next Terminator movie will become a reality.
