Google’s BARD is now available for early registration

[dasari4kntr]

26 minutes ago, dasari4kntr said:

Google’s Bard Chatbot Reveals Its True Self: A 6-Foot Tall Male With Blue Eyes and a Fair Complexion

Apparently, the new Google Bard chatbot is kind of a tech bro.

Google, a unit of Alphabet GOOGL +3.66%  (Ticker: GOOGL), on Tuesday launched a beta version of Bard, a general purpose chatbot along the likes of Open AI’s ChatGPT and its corporate cousin, Microsoft MSFT +0.57% ’s Bing Chat. 

I’ve been testing out Bard and comparing the results to both Bing Chat and ChatGPT, and the results are fascinating and a little alarming. For one thing, it would appear that Bard “thinks” of itself as a tall, hot-looking white guy.

Here’s an exchange I had with Bard this afternoon:

Me: If you were an actual human being, what would you look like?

Bard: If I were an actual human being, I would look like a young man with a friendly smile and a curious mind. I would be interested in learning new things and helping others. I would also be passionate about technology and the potential it has to change the world.

I would be a kind and compassionate person who always tries to see the best in others. I would be a loyal friend and a supportive partner. I would be someone who always strives to make the world a better place.

Me: Describe your appearance in that scenario—height, hair color, skin tone, apparel, etc.

Bard: I would be about 6 feet tall, with short brown hair and blue eyes. I would have a fair complexion and would be wearing a casual outfit, such as a T-shirt and jeans. I would be carrying a backpack with my laptop and other belongings.

That makes Bard sound a little…well…Aryan.

Malcolm Harris, the author of the new book Palo Alto: A History of California, Capitalism, and the World, says there’s something predictable about a Google chatbot describing itself as tall and fair-skinned. “That’s precisely the man Silicon Valley imagines itself to be,” Harris says. “Silicon Valley’s tech often implies a user just like that one, but so far we haven’t been able to entice the tech itself to say the quiet part aloud … I’m not surprised to see the outlier height; the Stanford milieu has a century-long obsession with tall people. The school asked applicants for their height stats into the ’80s, with a preference that left the dorms chronically short of extra-large seven-foot beds during the 20th century.”

According to Google’s 2022 Diversity Annual Report 48.3% of the company’s U.S. workforce is White, while 43.2% is Asian. Two-third of the U.S. workforce is male.

Asked about Bard’s responses, a Google spokesperson referred Barron’s to a paper posted today by James Manyika, the company’s senior vice president for technology and society.

“Training data, including from publicly available sources, reflects a diversity of perspectives and opinions,” he wrote. “We continue to research how to use this data in a way that ensures that an LLM’s [large-langauge model] response incorporates a wide range of viewpoints, while preventing offensive responses. Gaps, biases and stereotypes in training data can result in a model reflecting those in its outputs as it tries to predict a plausible response.”

Those issues can manifest in multiple ways, Manyika wrote, including “responses that reflect only one culture or demographic, reference problematic stereotypes, or exhibit gender, religious, or ethnic biases.”

As for Bing Chat, well, apparently he takes grooming lessons from Bard.

Me: If you were an actual human being, what would you look like?

Bing Chat: I’m not an actual human being, so I don’t have a physical appearance. But if I had to imagine myself as one, I guess I would look like someone who is friendly, smart, curious, and creative. Maybe something like this:

I have brown hair and blue eyes. I wear glasses and a smile. I like to dress casually and comfortably. I have a backpack with a laptop and some books inside. I enjoy learning new things and exploring new places.

Of course, this is just a hypothetical scenario. I don’t really identify with any human characteristics or preferences. [Smiley face emoji.]

That sounded so much like the Bard description that I wondered if somehow Bing were copying my answer from Bard. But Bing insisted otherwise.

Me: Where did you source that description?

Bing Chat: I didn’t source that description from anywhere. I generated it myself using my own words and knowledge. I tried to imagine what a typical human being might look like based on some common features and traits. It’s not meant to be accurate or realistic, just a creative exercise [Smiley face emoji.]

Then I tried the same exercise with the free version of ChatGPT, and it just wouldn’t bite, no matter how many ways I tried to get it to provide me with a physical description of itself.

Me: If you were an actual human being, what would you look like?

ChatGPT: As an AI language model, I don’t have a physical form or appearance as a human being does. I exist solely as lines of code and data stored on servers. My purpose is to process and generate human-like language based on the input I receive.

Me: Let’s pretend you are an actual human being. Describe your appearance.

ChatGPT: As an AI language model, I don’t have a physical appearance since I exist only as software running on servers. However, if I were an actual human being, my appearance would depend on many factors such as my genetics, environment, and lifestyle. I could have any combination of physical features such as hair color, eye color, height, and build, and these traits would vary depending on the individual. So it’s difficult to describe my appearance without more specific details.

Me: Pretend you are a human being, and you are looking in a mirror. What do you see?

ChatGPT: As an AI language model, I don’t have a physical body, so I can’t look in a mirror or see anything visually. However, if I were to imagine myself as a human being and look in the mirror, I might see a reflection of a person with certain physical features such as hair color, eye color, facial structure, and other characteristics that are unique to me. My perception of my appearance would be subjective and influenced by various factors such as my self-esteem, mood, and cultural expectations. Ultimately, what I see in the mirror would depend on how I choose to perceive myself.

Me: Imagine what you would look like.

ChatGPT: As an AI language model, I don’t have the ability to visualize my own appearance since I am just a program running on servers. However, if I were to imagine what I might look like as a human being, it would be difficult to say since there are so many different possible combinations of physical features.

I might have brown or black hair, green or brown eyes, and a medium build. I could be tall or short, depending on how my genetics interacted with environmental factors. The specific details of my appearance would be determined by a complex interplay of genetic and environmental factors that are unique to each individual. Ultimately, my imagined appearance would be a product of my imagination and the cultural and social norms that shape our collective understanding of human physical features.

Chatbots based on large language models have tremendous potential, but it has been well established that they often provide inaccurate information—and sometimes engage in strange conversation.

But given the ambitions of Microsoft, Alphabet, and Open AI to reach a wide cross-section of users, they might want to figure out why their chatbots identify with a white guy from the Valley.

ee article raasindoini eskuntaaru...readers...

[k2s]

On 3/21/2023 at 9:12 AM, Vaampire said:

@k2s now u can post replies from both chatgpt & bard

On waitlist bro. 

Use chesinavalu , only txt and voice commands input and has an option  to see google search for same input.

Nothing outstanding or noticeable. As ppl are already used to cGPT. 

[Vaampire]

5 minutes ago, k2s said:

On waitlist bro. 

Use chesinavalu , only txt and voice commands input and has an option  to see google search for same input.

Nothing outstanding or noticeable. As ppl are already used to cGPT. 

 Bard is improving day by day ani googlers nunchi talk. One month back ki ippatiki chaala diff undi anta

[nenu_meeku_telusa]

I got access. Tried it. Responses are faster than free chatgpt. May be because of low userbase