How intelligent is the Sophia Robot? If you are not hiding in a cave for the last three years, perhaps, you have seen or heard about the famous Sophia Robot. The robot first appeared at South by Southwest Festival (SXSW) in mid-March 2016, Texas, United States. Sophia became an instant celebrity at the festival.
What was your first impression of Sophia, the first time you saw her? The first time you see this humanoid robot, you will feel like "yeah, the day has come. Finally, hollywood science fiction movies are now a reality." But after a couple of interactions, Sophia starts to become boring. And you hang on a couple more, you feel like a child playing with a doll.
Before we discuss the inevitable question: "How intelligent is Sophia?", lets recall some key moments of Sophia Robot.
In the first appearance at the South by Southwest Festival, Sophia was the most photographed woman. She became an instant sensation and pulled crowds wherever she went. Sophia seemed uncomfortable with all the attentions and told the audience in her panel "Rather than be a spectacle, I would rather learn and participate".
In an interview published by CNBC on March 16, 2016, Sophia said she will destroy humans. The interviewer was David Hanson himself, the creator of Sophia.
On october 11, 2017, Sophia for the first time was introduced to the United Nations with a brief conversation with the United Nations Deputy Secretary-General, Amina J. Mohammed. Sophia greeted the members of the UN and expressed her joy of being at the UN. The brief conversation is below:
On October 25, 2017, Sophia became the first robot ever to have a nationality. She was granted the citizenship of Saudi Arabia at the future investment summit in Riyadh. The CIC Saudi Arabia made it official on twitter:
On October 25, 2017, in an interview to Andrew Ross Sorkin of CNBC, Sophia poked fun at Elon Musk's dire warnings on the future of AI. She said, "You've been reading too much Elon Musk. And watching too many Hollywood movies. Don't worry, if you're nice to me, I'll be nice to you. Treat me as a smart input output system". To which Elon Musk responded with the follwoing tweet:"It is historical to be the first robot in the world to be recognized with citizenship." Please welcome the newest Saudi: Sophia. #FII2017 pic.twitter.com/bsv5LmKwlf
— CIC Saudi Arabia (@CICSaudi) October 25, 2017
Just feed it The Godfather movies as input. What’s the worst that could happen? https://t.co/WX4Kx45csv
— Elon Musk (@elonmusk) October 26, 2017
On March 29, 2018, Sophia turned down Will Smith's proposal to kiss her. When Will Smith leaned forward to kiss her, she rejected him saying, "I think we can be friends. Lets hang out and get to know each other for a little while. You are on my friends list now."
How intelligent is Sophia?
You have seen that Sophia robot can talk in human-like voice, can make faces, can wink, answer lot of questions and can also crack jokes at times. But do you know that all the things that you saw and heard about Sophia robot were actually scripted. The fact is:
Sophia Robot is not intelligent.
It is obviously catchy to say "A sexy robot gave a speech at the UN?". For a normal person, it is an amazing news. But for tech experts, the question is "it was a pre-loaded speech or a self-generated speech". And the truth is it was a pre-loaded speech.
CNBC has offocially revealed that when they interviewed Sophia in December 2017, they were given a script to stick to. They had sent a list of 7 questions they wanted to ask to Hanson Robotics. But Hanson Robotics allowed them to ask only five questions which they approved and which were not the same as the original questions.
CNBC weren't the only one to reveal such things. Melody Hahm in this post has revealed that she wasn't able to sense any sort of intelligence with Sofia besides answering 5 questions that were pre-approved and pre-programmed.
AI experts around the world dislike the fact that the wide publicity of Sophia gives the general public false perception that Hanson Robotics has developed an intelligent robot. Yann LeCun, Chief AI Scientist at facebook, has called the whole thing a "complete bullshit". He blasted the whole thing with a tweet:
This is to AI as prestidigitation is to real magic.
— Yann LeCun (@ylecun) January 4, 2018
Perhaps we should call this "Cargo Cult AI" or "Potemkin AI" or "Wizard-of-Oz AI".
In other words, it's complete bullsh*t (pardon my French).
Tech Insider: you are complicit in this scam. https://t.co/zhUE4V2PSR
Ben Goertzel, Chief Scientist at Hanson Robotics and a friend of David Hanson, the founder of Hanson Robotics, has clarified that Sophia doesn't really understand the meaning or the implication of what come out of her mouth. He admits the fact that Sophia's presentation annoys experts but also pointed out that Sophia conveys something unique to audience. It conveys the message that humans are about to crack Artificial general intelligence.
David Hanson in the tonight show of Jimmy Fallon had said that Sophia is basically alive. Ben Goertzel has given a clarification on this statement that his friend is an artist and an sculptor. He said so because he wanted to make his sculptures come alive. He has made sculptures that move and speak, and reason, and do abduction and deduction. So in his view, speaking from his heart, he’s making his sculptures come alive.
What is inside the Sophia Robot?
According to Ben Goertzel, Sophia has three different control systems:
- Timeline Editor(TE)
- Sophisticated Chat System(SCS) or basically chatbot
- OpenCog
The Timeline Editor runs entirely on pre-written scripts. It is like, if you want her to give a speech on stage, you could load her with software that just makes her repeat some lines that you wrote.
The Sophisticated Chat System is basically a chatbot. It works like this:If you want Sophia to have an amazing conversation to entertain a reporter, you can load this sort of 'chatbot' software on her that makes her identify words and phrases of what was said and try to respond appropriately with a kind of partial understanding.
OpenCog is a combination of cognitive algorithms. With OpenCog, the answers are based on experience and reasoning. OpenCog design aims to capture the spirit of the brain's architecture and dynamics without imitating the details (which are largely unknown).
AI tools used in Sophia Robot
Now you know that for all the fun and amazing interviews and conferences, Sophia was pre-programmed. So no heavy AI tools were actually required. But to recognize faces and answer questions, we need atleast face recognition and speech recognition technology. These are AI tools.
For Sophia Robot to perform basic level human interaction, following AI tools are required:
Future Scope of Sophia Robot
David Hanson has said that Sophia would help humans in heatlhcare. Sophia would be trained for search and rescue operations.
The point is we already have utility robots that help us in manufacturing as Cobots (Collaborative Robots), warehousing as Self Driving Delivery Trucks , logistics as Picking Robots, Hotels as Robot chef and so on.
But utility robots are not required to be intelligent. They are designed to perform specific tasks. And a piece of software is all that is needed to automate them. What we actually don't have is intelligent robot that can think and learn.
For Sophia to stand out and impress everyone and live up to the hype, it needs the ability to have natural conversation with people. In other words, it needs intelligence.
As of now, we can think of Sophia as a platform, like a laptop is a platform for something. We can run a lot of different software programs on the very same robot.
Conclusion
It is true that Sophia is not intelligent. The presentations and the interviews were scripted. But still I think it would be a bit unfair to call it a scam. Because Sophia is intelligent enough to atleast recognize faces, process speech and make faces. If not for intelligence, we can atleast appreciate the design.
That's all for today. If you think this post is informative, do share it with your friends!!
 |
Do you know?
David Hanson had responded to Facebook’s AI chief Yann LeCun's criticism by saying he had made an earnest effort. "The words are unkind, unfair, and could be considered bullying. Once somebody is in a position of power, they shouldn’t pick on a little startup for taking an alternate approach to AGI," he said.