In: Computer Science
just anwser .the question, so that I can finish the essay
[10 marks] Given the following article in Computer Technology:
We can’t trust AI systems built on deep learning alone
Gary Marcus, a leader in the field, discusses how we could achieve general intelligence— and why that might make machines safer.
by Karen Hao Sep 27, 2019
https://www.technologyreview.com/s/614443/we-cant-trust-ai-systems-built-on-deeplearning-alone/
INSTRUCTIONS: Answer the following questions in an essay:
1. What are the value and descriptive assumptions? [4 marks]
2. Is/Are there any fallacy/fallacies in the reasoning? [4 marks]
3. What kind of evidence is provided and how good is it? [2 marks]
1. The author's value assumption is that General AI is much superior to narrow AI when it comes to solving a real-world problem.
The author's descriptive assumption is that the world is using giving too much attention and credit to narrow AI. Currently, different kids of cognitive solutions that require different solutions are being obtained by the same type of approach (narrow AI).
2. I agree with the author. If we are going to use an AI to solve a real-world problem, it should be aware of all the outliers in the problem(and there are a lot of outliers when it comes to real-world problems like self-driving cars). It is risky to use a narrow AI to solve such problems.
3. The speaker gives examples like using a picture of an elephant in a silhouette for a recognition system, challenging AlphaGO in a rectangular board to prove that the current artificial intelligence is specific to a particular task only. He also gives examples like hate speech detectors that are easy to fool, job application systems that perpetuate discrimination, and self-driving cars that have crashed, sometimes killing the driver or a pedestrian, to prove that even the task it's assigned to do is not being done in a desired and safe way. This evidence is good to support his assumption.