Myths and realities about AI
Mildred López
Yesterday, as I was preparing my daily reflection about Artificial Intelligence (AI), I did a simple search in Google for the prompt “artificial intelligence is….”. I’ll share with you a screenshot of the 10 first predictions that the search bar got me:
These represent common myths or assumptions about AI. I would love for us to reflect together; therefore, I’ll share a video for each and my comments:
1. Artificial Intelligence is not prone to any bias or discrimination.
The reality is that it is prone to bias. The first step is to recognize them and build systems to address them. It is imperative that we understand how algorithms work and critically use the information it provides.
See Jabril’s collab with CrashCourse shares 5 types of bias:
- Data reflects existing biases.
- Unbalanced classes in training data.
- Data doesn’t capture the right value.
- Data amplified by a feedback loop.
- Malicious data attack or manipulation.
https://www.youtube.com/watch?v=gV0_raKR2UQ
2. Artificial Intelligence is dangerous.
Define dangerous. I want to share with you the analysis of Joanna Stern in the Wall Street Journal. She used several challenges to see if AI with pass as a human. In several cases, it already passed:
https://www.youtube.com/watch?v=t52Bi-ZUZjA
3. Artificial Intelligence is going to take over our homes.
I am convinced it will get there, especially in monotonous tasks that can be automated. Tilman Wolff and Ranga Yogeshwar documentary for DW Documentary explains different applications, its pros, and cons:
https://www.youtube.com/watch?v=s0dMTAQM4cw
4. Artificial Intelligence is about.
Listen to the excerpts of Stuart Rusell on Radio Davos:
https://www.youtube.com/watch?v=RzkD_rTEBYs
5. Artificial Intelligence is a threat to humanity debate.
Beyond sci-fi movies, we need ground rules to work. Previous development rules for AI are no longer in place. Mo Gawdat (former Chief Business Officer for Google X) discusses with Piers Morgan the speed at which AI is learning. He mentioned 3 boundaries that they had for AI:
- don’t put it on the open internet until the control problem is solved,
- don’t teach it to code because that makes it self-develop,
- don’t have other agents prompting it.
I can’t forget one line of the interview where they mentioned, “I am not concerned of the machines…I am concerned of humans with AI in their hands”. You can see the full interview here:
https://www.youtube.com/watch?v=oxRZqzth9r4
We need to focus not only on the rules for development but also on the implementation rules of these platforms.
6. Artificial Intelligence is about mcq.
This prompt represents questions, multiple-choice questions about AI. It is a great opportunity to test our understanding of the topic. I’ll like to share a WIRED video of Gary Marcus, an expert on AI, answering questions from Twitter:
https://www.youtube.com/watch?v=Puo3VkPkNZ4
7. Artificial Intelligence is associated with computers of which generation.
The concept of generations of computers is explained briefly by Simply Coding at:
https://www.youtube.com/watch?v=NqgpZ_v4Ne8
8. Artificial Intelligence issues.
The Center for Humane Technology presents Tristan Harris and Aza Raskin, as they discuss the dilemma of AI development. See the full video at:
https://www.youtube.com/watch?v=xoVJKj8lcNQ
9. Artificial Intelligence is good or bad.
We cannot categorize it as good or bad. AI will work to do what we ask it to do, therefore, it depends on who is using it. See Janelle Shane TED conference on the danger of AI is weird than you think. She presents several examples for discussion.
https://www.youtube.com/watch?v=OhCzX0iLnOc
10. Artificial Intelligence is defined as.
I enjoyed watching this video by Duke University; it leverages our understanding of pop culture, such as movies and series, to explain a complex concept in a relatable way:
https://www.youtube.com/watch?v=c0m6yaGlZh4
What other myths and realities do you know about AI?
No comments:
Post a Comment