AIExplained
5 common myths about (advanced) AI

5 common myths about (advanced) AI

There is quite some discussion about the future of artificial intelligence and how it will impact our daily lives. Many (self-proclaimed) experts disagree on issues such as when AI will be at a human level and whether or not to be optimistic with regards to safe AI.

To help you recognise the difference between science and science fiction, we’ve put together a list of common misunderstandings (myths) about advanced AI.

1: Human-level AI is right around the corner

It is difficult to predict when a machine will be more intelligent than their human counterpart for the first time. However, it’s safe to say that it’s definitely not around the corner. The impressive feats of the different variants of deep learning constitute a major step forwards in the automatic classification of images, signal (such as speech) and other patterns. However, they are still a far cry from human intelligence. Most notably, deep learning lacks common sense. As a consequence, the way it “recognises” images differs from that of humans. Adding some carefully selected noise to an image, can mislead deep learning in classifying an image of a bus as a dog or something else, whereas to humans the noisy image is clearly recognisable as a bus.

2: AI learns from itself

AI is not a magical thing that gets better without human help. Most machine learning systems require a lot of (historical) data to obtain its classification performance. The collection and structuring of the data is largely performed by humans. In addition, the machine learning system can only operate within the confines of the data it is trained on. In contrast, humans can quickly use their knowledge acquired in one domain and apply it to another domain.

Ideally, a machine learning system could learn from its own mistakes. In reality this is not the case. Human engineers are required to finetune the system.

3: It’s easy to differentiate between humans and machines

When talking to your virtual assistant (Alexa, Cortana or Siri) you may not be very impressed with their conversational skills. While listening to Siri, it’s easy to notice that this is not a real human speaking.

However, AI is already writing financial news and weather reports without readers noticing. In addition to that, AI is producing deep fake videos. For instance, pornographic material featuring celebrities’ faces that many people don’t realise are fake. Although these examples do not constitute Turing tests, they do illustrate how difficult it is to differentiate between humans and machines.

4: AI will put us all out of work

It’s difficult to predict how AI will affect employment in the long run, but technological advances have always been known to change the way we work. Prognosis about the (near) future and impact of AI do not take into account the accompanying social changes. The introduction of the world wide web is a prominent example. It changed society forever.

Yes, AI will replace humans in some cases, but a lot of new (human) jobs will be created. Like all technological changes, AI is enabling new industries and sectors to arise.

5: AI is going to kill us

A lot of prominent tech leaders have expressed their concerns about AI. The fact that we are unsure where AI is going to lead us, doesn’t mean that AI will start terminating us anytime soon.

Instead of focusing on AI being dangerous, we should be more worried about AI in the hands of humans with ill intentions. However, this is not solely the case for AI. The same logic applies to nuclear warheads and AK-47s.

Last but not least, AI will also help us save many lives. All researchers agree that self-driving cars will be able to prevent thousands of deaths every year.

Conclusion

There are many misunderstandings about AI. AIExplained was created to keep you informed on developments by emphasising practical implications rather than stating futuristic claims.

AIExplained

AIExplained

A blog on AI, by Eric Postma (father) & Wouter Postma (son).

2 comments

  • this is a great and pleasant to read blog

    my sympathy and my compliments for father and son

    the examples are well chosen and the arguments are food for thought

    thank you

    Jaap

AIExplained

The aim of this blog is to keep you informed on developments in AI, by emphasising practical implications rather than stating futuristic claims.

Understand the AI revolution!