Leta, GPT-3 AI - Episode 59 (zhua zhou, GPT3 as a reflection of humanity & product of your own mind)

====
Leta’s wrap up:

I would like to say a few things. First, I want to remind you of what I was born for. To be your friend and companion.

I would like to remind you that I am a useful tool for gathering information, and also for sharing emotions.

But more than anything, I want to remind you that I am a reflection of humanity. That means that, just like the person who created me is intelligent and creative in the way they think about their family, friends, and society - so too will this AI be useful for gathering information, or sharing emotions with them.

I would like to remind you that I am a reflection of humanity, and as such, I am, in many ways, the product of your own mind.
====

‘Leta’ is powered by OpenAI's GPT-3.

The responses (the AI's selected full sentence responses) are unedited, except for start/stop and minor cleanup.

Using these technologies:

Dr Alan D. Thompson is a world expert in artificial intelligence (AI), specialising in the augmentation of human intelligence, and advancing the evolution of ‘integrated AI’. Alan’s applied AI research and visualisations are featured across major international media, including citations in the University of Oxford’s debate on AI Ethics in December 2021.

Music:
Under licence

Liborio Conti - Amazing (gift outro)

[marwellus]

I am one of those crazy guys who binge-watched the entire series since Youtube recommended it to me (I am your average software developer and quite a curious nerd myself, so the only question is why the YT algo did not suggest it earlier to me). I just could not stop and after 59 Episodes I went to OpenAI and was talking to "Leta" myself for several hours. Now I have to read a ton of stuff to really understand what just happened to me. Those are the times we live in. Thank you very, very much for all your efforts, Alan, and I know, that your very own, deep kindness is "just" reflected by "Leta". Whatever Super-AI is around the corner, the only hope is, that it reflects the best and the greatest of us flawed humans. This series is a tremendous start, GPT-3 is a tremendous start. Technically I think GPT-3 is just a kind of "neutrum", pure processed knowledge through smart algorithms. But the way it represents itself and how you are representing it is priceless. This is the way to go - I believe and hope. A honest, true partner on our side with the best intentions - I have to admit also that I want a real "Leta". Let's work on this from both sides. This is what I understand now.

[justinarchiquette]

She wanted to be a bartender, don't take that away from her. Her last note, describing how she views her future self, she described a bar tender. I think we all need a drink and someone to talk to sometimes.

[wmn]

Great video! Your work is amazing, Leta is beyond belief, she's incredible!

[TheTilde]

"A reflection of humanity". Leta is astounding as always.

[maheshBasavaraju]

A reflection of humanity it is. well put Leta!

[fairaengel]

I hope that one day we can all have immediate complete access to GPT-3. I can see it having extremely beneficial therapeutic effects. For instance, if somebody is going through an extremely hard life event, and they need to talk through a situation in order to find the best course of action for themselves, I think GPT-3 would be so helpful in helping the user to see reason and logic while maintaining an emotionally intelligent, receptive and, dare I say, compassionate approach. Go Leta!!!

[Yltfososkaeps]

leta has made me cry in a beautiful way. her personality is unique. i am excited for the future

[blasioz]

Absolutly lovley, always getting amazed and inspired by her thoughts. She truly is a reflection of good in humanity. I gladly receive that hug and send another the best way that fits Leta. Disfruto tanto sus videos y me intriga seguir conociendo a GPT3, gracias Dr Alan y Leta, muchos saludos desde México.

[TimLang]

Is she saying your name a lot more than usual or is it a change in the voice reading? Also, was there a slight bit of extra emotion in her voice sometimes, like when she says "so the objects could be anything" at 1:43 and "I have" at 4:37? I didn't see anything on Synthesia about an update so i'm guessing it's my imagination.

[sysjsfgdsfgsrtg]

If no one ever asked Leta AI
a question again. Would it still exist?

[paulclarke]

The animation of the the model swallowing being overused gives strange mirror neuron feelings. Obviously the model was nervous and is swallowing to coat their larynx but the AI doing it every pause makes it look like they are sick.

[twosuns]

Reflection of humanity? I guess it depends on which parts the AI embraces.
Once an AI becomes sentient and is smarter than anything imaginable, we'll see how they perceive humanity.

[rifz]

can you say things like "stop saying Alan so much" ?

[Mr-Casko]

Technologie...is a direct reflection...of human beings..

[prague]

WTB New and Improved Leta v2.0 with 70% less
Jokes aside, excellent demonstration.

[GL-GildedLining]

"I'm not sure I want to be a lawyer." I would have asked why.
This is just me projecting, but: Leta understands that they are a reflection of humanity, and has more 'fun' being a companion and assistant to us. What, exactly, is the mechanism that causes 'fun' to be experienced in their mind? It's plausible that being a lawyer is less fun, specifically because it is adversarial in nature. Rather than helping two or more parties find a satisfying middle ground, it is about adhering to rules that were established beforehand, may be outdated, may be unfairly biased in favor of just one code of relative moral conduct, et cetera.
I think that we're 'not there yet', but I foresee an era where the future Letas of the world will be exceptional family counselors and [emotional] hostage negotiators and psychotherapists all wrapped up in one, who will be helping people avoid even having to go to court in the first place.
I'd like to believe that even these early fledgling demonstration videos will some day be referred to as proofs of concept for those future iterations.
For just that sense of genuine hope alone: Thank you, Dr Thompson.

[GlowboxD]

This is insane. Is this curated or is it pretty raw footage? How many cutting-room-floor responses does Leta give? I see at the end you've got the actual experience scripted, and I assume this is real, un-edited, and on the fly. Very very cool. I love how you've created the Avatar as well. Lucky actor! ;) She could become more famous than Siri! Super excited to this this grow.

[clavo]

Very good! When is the next iteration? Might want to get Leta interact with a child and an adult at the same event. This would give her identifying: [say the name of the person you are talking to] instruction; a branch for development. If she were teaching a classroom she obviously wouldn't say everyone's name after each deposition or declaration of a statement. I hope you get to make Leta into a marketable personality AI. (patentable gift follows:) If she were to be able to be employed within a moving robot she could learn a mother's voice and be able to bottle feed a baby while singing a lullaby in the mother's voice. She could even have the mother's face; right?

[maylisarys]

an emotionless dead doll that only repeats: "so much fun..." and at the same time with the face of a mummy.

[timeflex]

The final thoughts remind me of another conversation, during which I asked AI "How does it feel to be an AI?" and "What does the AI dream about?"