What comes to mind when you think of deepfakes? A report by CB Insights got me thinking the other day about deepfakes and their impact on AI (artificial intelligence), quantum, and more. In case you didn’t know, deepfakes combine the expressions deep learning and fake and artificial intelligence; and that’s what we’re talking about with next-gen hack tactics using AI.
There are a lot of market numbers about AI-as-a-Service market, AI in financial services, AI in the medical sector, AI in the automotive market, AI in marketing, and AI at the edge. There’s just so much to discuss when it comes to AI, and we talk about it relatively frequently in an attempt to try to cover it from all sides.
So, for instance, AI is increasingly being used for predictive analysis, and that’s boosting the demand for AI-as-a-Service. The global AI-as-a-Service market is supposed to generate more than $15 billion worth of value between now and 2024.
AI is also improving the service offerings for financial companies, which are using technologies like deep learning, natural language processing, and chatbots to improve customer service and provide more personalized financial services.
In the medical sector, there was a report that defined 2020 as a year of historic market value for AI in medicine, and it predicts growth between now and 2026. I found a similar report making almost the same exact claims for AI in automotive.
And, of course, in manufacturing, AI-powered technologies are being used to increase labor productivity substantially. Research also claims AI can increase productivity up to 40%.
AI can also perfect visual inspections in manufacturing, and it can unlock predictive maintenance. One of the key points in the CB Insights’ report is the potential impact of deepfakes in AI and the industries that use it.
Thus, deepfakes, if you’re not already familiar, are “hyper-realistic AI-generated images and videos.” The pros of deepfakes in industries like retail could mean swapping in faces so customers can see themselves interacting with products virtually.
Another kind of creepier use of deepfakes are third parties using influential people’s images and existing videos and altering them to appear to say or do things they didn’t really say or do.
What comes to mind, with super Tuesday behind us, one can’t help but think of the potential impact that convincing deepfakes could have on important events like presidential elections. What if people start creating deepfakes of candidates saying outrageous things that then get circulated on social media as if they were real?
The firm predicts that hybrid models that combine classical machine-learning algorithms with quantum AI will see practical applications in the near future.
One potential example I can think of would be improved fraud detection in the financial sector. Can you imagine the combination of AI and quantum computing for tasks that require pattern recognition and anomaly detection?
It could create a huge advantage. And I want to highlight one more point from this report, so let’s discuss next-gen hacking. CB Insights’ research puts forward the idea that next-gen hacks will evolve on two fronts: fooling AI systems and leveraging AI to launch sophisticated attacks.
The first, fooling AI systems, can be accomplished by actors with malicious intent exploiting inherent bias in an AI model. Data poisoning is another method cybercriminals can use to fool AI algorithms.
Basically, this means introducing something tiny that will trick AI into misinterpreting data, eventually creating issues and effectively corrupting the whole algorithm. The second idea, leveraging AI to launch sophisticated attacks, also needs to be on our radar.
AI-based malware, for instance, is a frightening concept, and we essentially need to be looking for ways to use AI against AI. You may remember IBM’s proof of concept, deeplocker, a set of evasive attack tools powered by AI. IBM developed deeplocker to try to understand how AI is being used or could be used to enhance malicious malware techniques.
It will be interesting to see if CB Insights’ report is correct if the hybrid models that combine classical machine-learning algorithms with quantum AI will truly see practical applications in the near future. If that’s correct AI will become more concrete and will move beyond hype to practical business value. How cool is that?
Want to tweet about this article? Use hashtags #IoT #AI #machinelearning #bigdata #cybersecurity #blockchain #digitaltransformation #cloud #edgecomputing #infrastructure #5G #sustainability #manufacturing #futureofwork