The Intelligence Hype
Artificial intelligence(AI) is the latest victim of the hype bubble. The concept of a sentient and self-aware machine which is the product of AI is too overrated and may not be the immediate future of current AI systems.
There are several reasons as to why this might be the case for at least the next couple of years, the major one being how the current AI systems learn. Most of the AI training, be it any algorithm, is data-driven and manual.
The intelligence of such models is as good as the data they are given and due to the lack of cognition, the decision making power of this AI is often logical and is based on past data points it has seen during the learning process. So they may never reach the levels of human intelligence which is based on reasoning and problem solving combined with cognition.
When an AI system tries to make a decision or arrive at a solution, it relies on its training on past numeric data to identify patterns and act accordingly, which means AI is another way to manage huge data to derive insights.
Am I saying it has no foreseeable future? No, absolutely not. AI has huge potential and will change the way we interact with machines and will drive future innovation. It’s just that, it will not be like the way these hypesters are predicting or forcing you to believe.
Remember Jian Yang’s smart refrigerator from the series Silicon Valley that was hacked by Gilfoyle. You are forgiven if you don’t remember and banished from this article if you have not seen the series. Just kidding!!! Anyway, that refrigerator was part of a network of intelligent refrigerators connected to each other in a distributed fashion and was sharing information and data amongst themselves.
This concept is a slight glimpse of what our future holds and will be powered by a new tech jargon — AIoT. It is just the combination of two technologies, AI and Internet of Things (IoT). With more connected devices, this information sharing will see new applications and will make AI insights accessible to the general consumer. AIoT offers newer ways of smart automation where data driven inference and decision making will be more real time and robust.
Gone are the days when AI or machine learning relied on huge computing resources. Newer AIoT edge hardwares will inherit neural processing capabilities to drive edge inference of complex neural networks or data workloads. Does that mean Skynet is imminent? Nope, not really.
You would know this for a fact if you have even remotely tried building a machine learning or AI model. The current technology stack or frameworks which let you build AI models are hardware centric and data driven. So its definitely going to take some time until you see artificial general intelligence (AGI) or self aware machines.
The same was mentioned in the rather controversial tweet by Facebook’s Head of AI, Jerome Pesenti, when he tweeted,
“I believe a lot of people in the AI community would be ok saying it publicly. @elonmusk has no idea what he is talking about when he talks about AI. There is no such thing as AGI and we are nowhere near matching human intelligence”.
This spurred a lot of debate among the online community and I am in no position to make any comments on it. I am not an advocate of Mr. Pesenti or Mr. Musk, but what I do know is that an AGI is not our immediate future ( but could be possible in the long run) and the whole “AGI takeover” discussion would cause a negative distraction, wherein we fail to concentrate on the real problems faced by AI.
Biggest one of which is data privacy and security. Tech giants like Google and Facebook are already ahead of the curve in terms of AI solutions dispatched to their consumer and when it comes to transparency in operation of such systems.
Lengthy privacy policy documents are no way to address these woes because to the general public, this document means squat. What is required is awareness, decentralization and open-sourcing. AI inherits a lot of other shortcomings as well, but this article is not focusing on those as of now.
AI is still in its infancy and I feel as technologist, myself and other fellow AI evangelists should support and nurture this technology by concentrating on the right issues and sharing the knowledge among everyone.
Discussions on how AI will impact us in the future, what are some common misinterpretations regarding AI, decentralizing AI etc are beyond the scope of this article. It has often bugged me how AI is being misunderstood by general public and the industry as a whole. So, I am planning to start a new series where I try to demystify common misconceptions about AI, give you real stories on building AI solutions, give you tips on building scalable machine intelligence systems and keep you updated on new edge AI trends. Do let me know what you think in the comments below.
