Sunday, April 2, 2023

Bard - Google's Version Of Chat GPT & My Thoughts On LLMs, AI et al.

I guess since I work in the field, Google last week invited me to experiment with Bard.

My thoughts: 

  • Bard is extremely faster than Chat GPT3. It's obvious since Google has faster servers and more dollars than Open AI. 
  • Bard is probably trained on much richer data than Open AI since Google has decades of "our" data. But yet, Brad is more cautious in its answers than Chat GPT. 
  • Training data is everything in the current state of AI. And also, training data is almost everything for both humans and non-human animals. The sooner we understand this, the better the chances of us solving the problems we face. 
  • I am naturally drawn toward people and models who use the phrase "I don't know". In those terms Chat GPT and Brad are more humble than most humans since these two models have more knowledge than any one human ever.
  • They don't have wisdom; wisdom is knowing when to follow and when to break rules (and much  more). To be clear, one should never use the word wisdom with AI (leave alone LLMs)
  • Both models are highly creative. Creativity has been seen in all non-human animals and now these models are exposing our empty pride. I feel humble and understand I am an organic matter better with these models.
  • I don't ask questions to make myself look smart since I have a sense of what they are good and bad at. 
  • I asked both models about my favorite human, Montaigne. 

Brad's answer: 

Montaigne had a very close relationship with his cat. He wrote about her in his essays, and she was often a source of comfort and companionship for him. Montaigne believed that animals were capable of feeling emotions, and he saw his cat as a friend and a fellow creature. He wrote about how she would follow him around the house, and how she would sleep on his bed at night. He also wrote about how she would play with him, and how she would sometimes get into trouble. Montaigne's relationship with his cat is an example of the deep bond that can exist between humans and animals.

Here is an excerpt from one of Montaigne's essays about his cat:

"I have a cat, who does not much care to be seen, and whom I seldom see myself. She comes and goes as she pleases, and I never trouble her. She lies at my feet when I am at work, and purrs so loudly that she sometimes interrupts my thoughts. But I am not angry with her for that, for I know that it is her way of showing her affection. She is a very clean cat, and takes great care of her fur. She washes herself several times a day, and always seems to be well pleased with herself. She is also a very playful cat, and loves to chase after her tail. She is a very good hunter, and often brings me mice and birds. I am very fond of my cat, and she is very fond of me. We are good friends, and I hope that we will always be together."

Chat GPT's answer: 

What was Montaigne relationship with his cat like?

Michel de Montaigne, the 16th-century French philosopher and essayist, had a very close relationship with his pet cat. In his famous essay "An Apology for Raymond Sebond," Montaigne wrote extensively about his observations of his feline companion and used his observations as a springboard for philosophical inquiry.

Montaigne was known to be a lover of animals, and his cat was a constant presence in his life. He observed the cat closely and was fascinated by its behavior, agility, and independent nature. He saw the cat as a fellow creature with its own unique perspective on the world, and he valued its company and companionship.

Montaigne's relationship with his cat was one of mutual respect and admiration. He marveled at the cat's ability to live in the present moment and to be content with its surroundings. He also recognized the cat's independence and self-sufficiency, which he saw as admirable qualities to emulate.

Overall, Montaigne's relationship with his cat was one of affection, admiration, and intellectual curiosity. His observations of his feline companion led him to reflect deeply on the nature of animal consciousness and the limits of human understanding.

Both answers are good and unique in their own way plus both answers are true. 

For a long time, AI had been monopolized by a handful of big companies. We should give Open AI credit for breaking that trend. 

We are going to immensely benefit when two companies (and more will be joining) competing to train better models. Capitalism has its flaws but this competition is a gift from Capitalism to us. 

Having said that, there will be a huge amount of positive externalities coming out of these models, and conversely, there will be some negative externalities. We cannot predict either one of those. 

The only risk the current version of AI can do will be because of bad people using it for malicious goals.
For centuries people have been using even mundane tools maliciously to back their religious, political and other ideologies. 

So understand - AI cannot be compared to giving a nuclear weapon to a sociopath. 

We have so many lessons learned from industrialization, wars, economics etc., which we can use to define some common sense rules and boundaries for AI. I will write more about this soon.

Both models write better English phrases than I do. But yet, this blog is a place for Max and I. I will never use these models to write a blog post. It will be in my own words until my last breath. 

No comments: