This question isn’t easily answered, and I am curious to see what other risks and uses that may arise in the future.
The field of AI is extremely broad and the concept itself is often vaguely defined. That being said, one thing is for sure, AI truly is a game changer. AI has been described as a ‘disruptive technology’, that is, a technology that displaces well-established products or technologies, creating whole new industries and markets. Some examples of disruptive technologies include steam power and the computer.
A quick glance at the potential uses of this technology makes this classification easy to understand. AI has the potential to reshape the world around us and it may come as a surprise to some how pervasive this technology already is in our modern, tech-filled lives. Let’s have a look at some of the ways that AI is currently in use, before getting into some more technical aspects of the technology.
Four main uses of AI in our lives right now:
Those are just a few of the ways we see AI in action today. We all know that computers can do amazing things, but those outside the field of computer science may struggle to understand how they do what they do. AI, in that regard, is no different.
AI can be broadly categorized into five categories:
The possibilities for the future use of these AI technologies is limitless. AI has computational power greater than the human brain by orders of magnitude. Efficiency could be greatly increased in transport networks, agriculture, medicine and countless other sectors of human activity.
What are the risks?
If you have ever watched a Terminator or Matrix film then you know about the risk of a singularity occurring. Basically, this is the point where an AI supersedes us humans as the most intelligent ‘being’ on earth. This is unlikely to occur and if it did, it would be far, far into the distant future.
Officials are less concerned about an apocalyptic AI takeover and more concerned about the practical risks associated with this technology. There are some serious risks that should be on the minds of legislators:
● Bias and discrimination: AI systems do not design themselves; humans design these tools. Humans also carry inherent biases and values which can make their way into the code. The Guardian investigated this problem in 2018, where they found that googling the phrase “unprofessional hairstyles for work” showed images of mainly black women with natural hair, but when they searched “professional hairstyles” it showed pictures of coiffed white women.
● Invasion of privacy: AI can threaten privacy both due to its design and its deployment. A massive amount of personal data is already collected by big tech companies. However, the use of AI in the space of ‘big data’ could create serious problems, where AI systems can target, profile, or nudge data owner subjects without their knowledge or consent.
● Unclear objectives and unintended outcomes: The World Economic Forum outlines this dilemma quite well. Imagine that you tell an AI system to eradicate cancer from the world. It does so, but only by eradicating humans. The system succeeded in its goal very efficiently, but not in the way humans intended it to. Consider also the moral dilemma with driverless cars. In the event of a crash, does the car kill the 4 older people inside the vehicle, or does it swerve and kill the 6 younger people on the sidewalk?
I'm sure there are many more risks and benefits associated with this disruptive technology. And I am very interested in hearing your opinions!
AI is surely a simulation of human intelligence in machines (similar to the android Data in Star trek next generation) who strives to be more human. I suppose, it is where they (or he) are programmed to think, behave like human beings and mimic their actions. It is also about us accepting new forms of life/intelligence.
Codie Rossi
,Reference [1] covers the definition of AI fairly well. Their adoption of rational agent as an umbrella approach to what seems disparate subjects is for the most part a common approach to the definition of AI.
Now on second part, the risks that you mention are mainly abuses or bad implementations on the part of the designer. There are also attacks directed directly at machine learning such as those in adversarial machine learning [2] and machine learning Trojans[3], compromises against sensors[4], and cyber physical attacks[5].
References
[1Russell, S., & Norvig, P. (2010). Artificial intelligence: a modern approach.
[2]Huang, L., Joseph, A. D., Nelson, B., Rubinstein, B. I., & Tygar, J. D. (2011, October). Adversarial machine learning. In Proceedings of the 4th ACM workshop on Security and artificial intelligence (pp. 43-58).
[3]Geigel, A. (2014). Unsupervised learning trojan.
[4] Tippenhauer, N. O., Pöpper, C., Rasmussen, K. B., & Capkun, S. (2011, October). On the requirements for successful GPS spoofing attacks. In Proceedings of the 18th ACM conference on Computer and communications security (pp. 75-86).
[5] Al Faruque, M. A., Chhetri, S. R., Canedo, A., & Wan, J. (2016, April). Acoustic side-channel attacks on additive manufacturing systems. In 2016 ACM/IEEE 7th international conference on Cyber-Physical Systems (ICCPS) (pp. 1-10). IEEE.
Please review and answer this question about Trolley problem in AVs
https://www.researchgate.net/post/How_do_you_discuss_the_various_opinions_on_Trolley_Problem_in_Automouns_Vehicles
Hi Codie Rossi
,Check these out we discuss similar concerns about AI
Chapter The dark side of artificial intelligence in retail services innovation
Chapter Like a Cog in a Machine: The Effectiveness of AI-Powered Hum...
Thanks, Ali
Ali B. Mahmoud ,
I read your paper "The dark side of artificial intelligence in retail services innovation" and I have the following observations on two threats identified in the paper
These two threats merit very careful attention so that the wrong conclusion is not derived from them.
Regards
- Badges
- Science topic
- Similar topics
- Psychology
- Cognitive Science
- Mind