Wednesday, March 13, 2024

Face of AI

I have the idea of connecting a face to an LLM that can offer real-time emotional expression to represent the system's reaction to verbal and visual input.

My original purpose is to provide realistic feedback to my presentation of content. 

A practical use case would be to provide analysis and evaluation while simulating emotional responses to storytelling, comedy, instruction, explanation, etc., thereby offering presenters a natural target for content development and providing a medium for practicing and refining a presentation.

The concept, however, approaches the challenge of simulating personality and raises the question of sensibility - i.e. - different personalities respond differently. What about simulating a crowd? What about an intoxicated crowd? Hecklers? Perhaps these variants may be filtered with settings, but the concept also raised the question of innate bias in LLMs, and the need for unfiltered or 'objective' LLMs.

Can an LLM have a philosophy, or an attitude, or sensibility?

Can an LLM recognize, comprehend or appreciate the humor of impressions or funny noises given a voice or computer vision input?

Can an LLM recognize and appreciate humorous delivery of a joke, for example, subtle sarcasm in the tone of voice or facial expression, regardless of the context of the words?

If an LLM had such capabilities, sentiment analysis, etc., could it laugh and how would it laugh? What would account for the manner of laugh? 

How might different LLMs have different 'senses of humor'?

For that matter, how might human laughter be analyzed and taxonomized by an AI system?

It's interesting to think of an LLM having involuntary responses, as if it had a parasympathetic nervous system.

Certainly, given a context, ChatGPT can analyze and evaluate, but can it express a reaction in real-time or does it wait until it receives the entire spiel? Is the LLM self aware of it's own bias?

I searched 'Does ChatGPT have a face' and found this article about a company called Zero Distance.

Here's their YouTube channel.

Here's the product website, WEHEAD.



Cognition AI launches Devin, AI software engineer

https://siliconangle.com/2024/03/12/cognition-launches-devin-generative-ai-powered-coding-engineer/

https://www.youtube.com/@Cognition-Labs

Monday, February 26, 2024

Scientists Create DVD-Sized Disk Storing 1 Petabit (125,000 Gigabytes) of Data

Scientists Create DVD-Sized Disk Storing 1 Petabit (125,000 Gigabytes) of Data (popsci.com)95

Popular Science points out that for encoding data, "optical disks almost always offer just a single, 2D layer — that reflective, silver underside."

"If you could boost a disk's number of available, encodable layers, however, you could hypothetically gain a massive amount of extra space..."Researchers at the University of Shanghai for Science and Technology recently set out to do just that, and published the results earlier this week in the journal, Nature. Using a 54-nanometer laser, the team managed to record a 100 layers of data onto an optical disk, with each tier separated by just 1 micrometer. The final result is an optical disk with a three-dimensional stack of data layers capable of holding a whopping 1 petabit (Pb) of information — that's equivalent to 125,000 gigabytes of data...

As Gizmodo offers for reference, that same petabit of information would require roughly a six-and-a-half foot tall stack of HHD drives — if you tried to encode the same amount of data onto Blu-rays, you'd need around 10,000 blank ones to complete your (extremely inefficient) challenge.

To pull off their accomplishment, engineers needed to create an entirely new material for their optical disk's film... AIE-DDPR film utilizes a combination of specialized, photosensitive molecules capable of absorbing photonic data at a nanoscale level, which is then encoded using a high-tech dual-laser array. Because AIE-DDPR is so incredibly transparent, designers could apply layer-upon-layer to an optical disk without worrying about degrading the overall data. This basically generated a 3D "box" for digitized information, thus exponentially raising the normal-sized disk's capacity.

Thanks to long-time Slashdot reader hackingbear for sharing the news. 

Bezos, Nvidia Join OpenAI in Funding Humanoid Robot Startup

 

https://www.figure.ai/

https://www.youtube.com/@figureai

https://www.youtube.com/watch?v=Sq1QZB5baNw&t=15s


Update: China's fastest humanoid:

https://www.youtube.com/@unitreerobotics

https://m.unitree.com/


Bezos, Nvidia Join OpenAI in Funding Humanoid Robot Startup (msn.com)6

OpenAI, Microsoft, Nvidia, and Jeff Bezos are all part of a pack of investors in a business "developing human-like robots," reports Bloomberg, "according to people with knowledge of the situation..."

At the startup — which is named "Figure" — engineers "are working on a robot that looks and moves like a human. The company has said it hopes its machine, called Figure 01, will be able to perform dangerous jobs that are unsuitable for people and that its technology will help alleviate labor shortages."Figure is raising about $675 million in a funding round that carries a pre-money valuation of roughly $2 billion, said the people, who asked not to be identified because the matter is private. Through his firm Explore Investments LLC, Bezos has committed $100 million. Microsoft is investing $95 million, while Nvidia and an Amazon.com Inc.-affiliated fund are each providing $50 million... Other technology companies are involved as well. Intel Corp.'s venture capital arm is pouring in $25 million, and LG Innotek is providing $8.5 million. Samsung's investment group, meanwhile, committed $5 million. Backers also include venture firms Parkway Venture Capital, which is investing $100 million, and Align Ventures, which is providing $90 million...

The AI robotics industry has been busy lately. Earlier this year, OpenAI-backed Norwegian robotics startup 1X Technologies AS raised $100 million. Vancouver-based Sanctuary AI is developing a humanoid robot called Phoenix. And Tesla Inc. is working on a robot called Optimus, with Elon Musk calling it one of his most important projects. Agility Robotics, which Amazon backed in 2022, has bots in testing at one of the retailer's warehouses.
Bloomberg calls the investments in Figure "part of a scramble to find new applications for artificial intelligence."

Sunday, February 25, 2024

High-sugar product ingredients printed using low contrast colors

 

After deciding to cut any foods with corn syrup from my diet, I've noticed many products print their ingredients with low-contrast colors, such as white text on light-colored backgrounds.

My awareness of web accessibility requirements further brings the problem to my attention and I suspect that companies are deliberately masking their ingredients to prevent buyers from avoiding high-sugar products. 

The simple solution is to list ingredients using black text. 

Arizona Tea may not encourage their customers to read the ingredients, but they sure as hell ensure scanners can read the bar code.

Today, I noticed Arizona Tea Fruit Snacks, below.

Reading the ingredients online, I see:

PEAR JUICE FROM FRUIT JUICE CONCENTRATE, GLUCOSE SYRUP, SUGAR, MODIFIED FOOD STARCH (CORN), GELATIN, DEXTROSE; CITRIC ACID, GREEN TEA EXTRACT; NATURAL FLAVOR, FRUIT & VEGETABLE JUICE FOR COLOR (ELDERBERRY, GRAPE, SPIRULINA, TURMERIC); COLOR (PAPRIKA OLEORESIN); CARNAUBA WAX.

With turmeric, spirulina, and the anti-oxidants from the green tea extract, I'd better double up on my Arizona Fruit Snacks to prevent cancer.

The product description refers to the snack as "guilt-free" on the basis of being fat and gluten-free - a similar argument listed on candy packaging such as candy corn and gummy worms (which I love):

"From your favorite beverage to your next healthier snack time staple. AriZona Green Tea Fruit Snacks gives you 100% real fruit and only 100 calories per serving in the perfect little package. Enjoy Original, Apple, Mandarin and Plum Blueberry in this bite-size, fat free, gluten free, preservative free and 100% guilt free treat!"

Online reference suggests glucose syrup and corn syrup have equally adverse health effects.

I'll likely add additional example photos to this entry.




Thursday, February 15, 2024

Amazon LLM's 'emergent Abilities'

 

Largest Text-To-Speech AI Model Yet Shows 'Emergent Abilities' (techcrunch.com)46

Devin Coldeway reports via TechCrunch:Researchers at Amazon have trained the largest ever text-to-speech model yet, which they claim exhibits "emergent" qualities improving its ability to speak even complex sentences naturally. The breakthrough could be what the technology needs to escape the uncanny valley. These models were always going to grow and improve, but the researchers specifically hoped to see the kind of leap in ability that we observed once language models got past a certain size. For reasons unknown to us, once LLMs grow past a certain point, they start being way more robust and versatile, able to perform tasks they weren't trained to. That is not to say they are gaining sentience or anything, just that past a certain point their performance on certain conversational AI tasks hockey sticks. The team at Amazon AGI -- no secret what they're aiming at -- thought the same might happen as text-to-speech models grew as well, and their research suggests this is in fact the case.

The new model is called Big Adaptive Streamable TTS with Emergent abilities, which they have contorted into the abbreviation BASE TTS. The largest version of the model uses 100,000 hours of public domain speech, 90% of which is in English, the remainder in German, Dutch and Spanish. At 980 million parameters, BASE-large appears to be the biggest model in this category. They also trained 400M- and 150M-parameter models based on 10,000 and 1,000 hours of audio respectively, for comparison -- the idea being, if one of these models shows emergent behaviors but another doesn't, you have a range for where those behaviors begin to emerge. As it turns out, the medium-sized model showed the jump in capability the team was looking for, not necessarily in ordinary speech quality (it is reviewed better but only by a couple points) but in the set of emergent abilities they observed and measured. Here are examples of tricky text mentioned in the paper:

- Compound nouns: The Beckhams decided to rent a charming stone-built quaint countryside holiday cottage.
- Emotions: "Oh my gosh! Are we really going to the Maldives? That's unbelievable!" Jennie squealed, bouncing on her toes with uncontained glee.
- Foreign words: "Mr. Henry, renowned for his mise en place, orchestrated a seven-course meal, each dish a piece de resistance.
- Paralinguistics (i.e. readable non-words): "Shh, Lucy, shhh, we mustn't wake your baby brother," Tom whispered, as they tiptoed past the nursery.
- Punctuations: She received an odd text from her brother: 'Emergency @ home; call ASAP! Mom & Dad are worried... #familymatters.'
- Questions: But the Brexit question remains: After all the trials and tribulations, will the ministers find the answers in time?
-Syntactic complexities: The movie that De Moya who was recently awarded the lifetime achievement award starred in 2022 was a box-office hit, despite the mixed reviews.
You can read more examples of these difficult texts being spoken naturally here.

AI chatbot data harvesting personal data

 I'm sure chat isn't much better.


Your AI Girlfriend Is a Data-Harvesting Horror Show (gizmodo.com)102

"A lot of that AI chatbots that you spend days talking to push hard on getting more and more private information from you," writes longtime Slashdot reader michelcultivo, sharing a report from Gizmodo.

"To be perfectly blunt, AI girlfriends and boyfriends are not your friends," says Misha Rykov, a Mozilla Researcher from the company's *Privacy Not Included project. "Although they are marketed as something that will enhance your mental health and well-being, they specialize in delivering dependency, loneliness, and toxicity, all while prying as much data as possible from you." Gizmodo reports:Mozilla dug into 11 different AI romance chatbots, including popular apps such as Replika, Chai, Romantic AI, EVA AI Chat Bot & Soulmate, and CrushOn.AI. Every single one earned the Privacy Not Included label, putting these chatbots among the worst categories of products Mozilla has ever reviewed. You've heard stories about data problems before, but according to Mozilla, AI girlfriends violate your privacy in "disturbing new ways." For example, CrushOn.AI collects details including information about sexual health, use of medication, and gender-affirming care. 90% of the apps may sell or share user data for targeted ads and other purposes, and more than half won't let you delete the data they collect. Security was also a problem. Only one app, Genesia AI Friend & Partner, met Mozilla's minimum security standards.

One of the more striking findings came when Mozilla counted the trackers in these apps, little bits of code that collect data and share them with other companies for advertising and other purposes. Mozilla found the AI girlfriend apps used an average of 2,663 trackers per minute, though that number was driven up by Romantic AI, which called a whopping 24,354 trackers in just one minute of using the app. The privacy mess is even more troubling because the apps actively encourage you to share details that are far more personal than the kind of thing you might enter into a typical app. EVA AI Chat Bot & Soulmate pushes users to "share all your secrets and desires," and specifically asks for photos and voice recordings. It's worth noting that EVA was the only chatbot that didn't get dinged for how it uses that data, though the app did have security issues. [...]