AI/Machine Learning, Bharat and Bhartiya IT Industry

[sanman]

Top 10 Indian Robotics Startups

https://www.youtube.com/watch?v=Zx50uW5qEYI

[sanman]

When looking for new materials for defense applications, we should consider this approach:

https://deepmind.google/discover/blog/m ... -learning/

[NRao]

Convert Hindi Speech to Text (Python Tutorial)

5:24 minutes

[sanman]

Is GPT-4 vulnerable to theft?

https://www.youtube.com/watch?v=HR2PUCAukbQ

[sanman]

IBM QS2
(IBM Quantum System 2)

You better watch this - the Quantum Era is here:

[bala]

PM Narendra Modi inaugurates Global Partnership on AI (GPAI) Summit, 2023
at Bharat Mandap

https://www.youtube.com/watch?v=KZohjfaJ4tg

PM Speech

https://www.youtube.com/watch?v=w3qYdqxajTw

[sanman]

Tesla unveils "Optimus Gen-2" upgraded version of android robot:

[bala]

In this YT, Lt. Gen Raj Shukla is urging India to concentrate on AI and ergo extend this to the armed forces.

Lieutenant General Raj Shukla, PVSM,YSM, SM, ADC is a retired General Officer of the Indian Army who served as the General Officer Commanding-in-Chief of the Army Training Command (GOC-in-C ARTRAC). Shukla has been a Research Fellow at the Institute for Defense Studies and Analyses and a visiting fellow at the Centre for Land Warfare Studies.

Lt. Gen Raj Shukla & Vaibhav Discuss Fall of Western Powers, Rise of China, Artificial Intelligence

https://www.youtube.com/watch?v=uevIOIvIWKg

Raj Shukla is asking GOI to induct some top AI experts into MoD. L1 bidding is a waste of time and needs to change. Currently, India has no LLM (large language model), whereas China and USA have hundreds. National Security is at stake. AI winners will dominate the world. India has to be serious and put money behind the endeavor, also quantum computing is another area that requires serious investment. When India is looking at #3 or #2 in the world it has to think big and plan big.

[sanman]

https://analyticsindiamag.com/indian-ai ... -autonomy/

[sanman]

Researchers from SETI (Search for Extraterrestrial Intelligence) have carried out an experiment to communicate with humpback whales using AI.
I wish India's marine biologists would also attempt such communications too. Who knows what we could learn from these other species?

[sanman]

Meet 'Coscientist,' Your AI Lab Partner

https://www.sciencedaily.com/releases/2 ... 012729.htm

Date: December 21, 2023
Source: National Science Foundation
Summary:
An artificial intelligence-driven system has autonomously learned about certain Nobel Prize-winning chemical reactions and designed a successful laboratory procedure to make them. The AI did so in just a few minutes and correctly on its first attempt. According to the authors, this is the first time that a non-organic intelligence planned, designed and executed this complex reaction that was invented by humans.

Here's something to ponder:

Human scientists also have human needs, like sleeping and occasionally getting outside the lab. Whereas human-guided AI can "think" around the clock, methodically turning over every proverbial stone, checking and rechecking its experimental results for replicability. "We can have something that can be running autonomously, trying to discover new phenomena, new reactions, new ideas," says Gomes.

"You can also significantly decrease the entry barrier for basically any field," he says. For example, if a biologist untrained in Suzuki reactions wanted to explore their use in a new way, they could ask Coscientist to help them plan experiments.

"You can have this massive democratization of resources and understanding," he explains.

[sanman]

Soon even housekeepers and servants could be replaced:



Build it yourself:

https://mobile-aloha.github.io/

https://docs.google.com/document/d/1_3y ... VM6Dc/edit

They're quoting $32,000 -- could Indians build it for less?

[sanman]

This thread is in the Science & Technology section -- but should we have a separate thread in the Military section?
(ie. should we have a separate thread specifically on the military uses of AI?)


https://www.cnbc.com/2024/01/16/openai- ... tools.html

OpenAI Quietly Removes Ban on Military Use of Its AI Tools

[sanman]

Facebook/Meta and OpenAI each developing next-gen AI:

[sanman]

Some claim that humans are merely the precursors for a grander superintelligent AI, and this AI will likewise build an even greater AI, which will build an even greater AI...

[sanman]

hi sanman, appreciate what you are doing, for the videos that you post, could you also put in a relevant timestamp, or summarise the video through bard; the following are video summaries from bard for the youtube videos that you have posted on this page:
<snip>

Following your advice, I just tried using Google Bard to summarize this video, but it couldn't even get the name of the announcer/host right.
Hosts's name is Rishab Gulati, but Google Bard is referring to him as Sreenivasan Iyer:

https://www.youtube.com/watch?v=ARJnRVlzpP8

The speaker of the video, Mr. Sreenivasan Iyer, is discussing the potential of Artificial Intelligence (AI) in India with a panel of guests. He believes that India has a trillion-dollar opportunity to become a leader in AI, but that the country needs to take some key steps to make this happen.

Here are the seven key points that Mr. Iyer makes in the video:

India needs to invest in supercomputing power. This is essential for training AI algorithms, which require massive amounts of data and processing power. India's current supercomputing capabilities are far behind other developed countries.
India needs to develop its own AI-specific semiconductor chips. These chips are more efficient at running AI algorithms than traditional CPUs.
India needs to create a large-scale data repository. AI algorithms need to be trained on massive amounts of data, and India needs to make sure that its own companies have access to this data.
India needs to set up an "AI Mitra" program to allow volunteers to contribute to AI training. This would help to address the data shortage and also provide valuable human oversight for AI algorithms.
India needs to create an "AI City" to attract global AI companies. This would create a hub for AI innovation and development in India.
India needs to make it easier for Indian companies to access government data. This data could be used to train AI algorithms for a variety of applications.
India needs to develop a system for anonymizing data. This is essential for protecting people's privacy while still allowing AI companies to access the data they need.
Mr. Iyer concludes the video by saying that AI is a trillion-dollar opportunity for India, but that the country needs to take action now to make sure that it does not miss out on this opportunity.

I hope this summary is helpful!

[ricky_v]

Following your advice, I just tried using Google Bard to summarize this video, but it couldn't even get the name of the announcer/host right.
Hosts's name is Rishab Gulati, but Google Bard is referring to him as Sreenivasan Iyer:

well that is odd, i recommended bard over bing-gpt as per my experience, bard is much more flexible in its summarising viewpoints than bing, back to bing-gpt then

[S_Madhukar]

Kohima uses 3D digital twin and various drones including some 3D printed for urban planning - surprised to know they had a GIS centre set up in early 2000s

https://youtu.be/m9gaRKb1bZg?si=yH59ZIx11cS0MC9G

[S_Madhukar]

well that is odd, i recommended bard over bing-gpt as per my experience, bard is much more flexible in its summarising viewpoints than bing, back to bing-gpt then

Not sure of the latest version but last year Bard had the worst rate of hallucinations among the top GPTs

https://www.medrxiv.org/content/10.1101 ... 23288265v1

[sanman]

India is Building its Own "sovereign AI". Here's Why

https://www.youtube.com/watch?v=pEolSTz2RnQ

Google Summary

This video talks about AI sovereignty, the idea that a country should have its own artificial intelligence. The speaker says that AI will become more important in the future and that it is important for countries to have their own AI so that they can control their data and develop their own AI products. India is working on building its own AI ecosystem and the government has released a draft data governance framework that would make it easier for Indian startups and researchers to access data. The speaker also mentions that it will be difficult for India to catch up to Western AI firms, but that it is important to try.

Here are some key points from the video:

AI sovereignty is the idea that a country should have its own artificial intelligence.
AI will become more important in the future and it is important for countries to have their own AI so that they can control their data and develop their own AI products.
India is working on building its own AI ecosystem and the government has released a draft data governance framework that would make it easier for Indian startups and researchers to access data.
It will be difficult for India to catch up to Western AI firms, but it is important to try.
I hope this summary is helpful!

[sanman]

Google Gemini Calls Modi a 'Fascist', But Declines to Similarly Label Zelensky

https://www.youtube.com/watch?v=rhvjMl5o3HA

[sanman]

"AI Cannot Replace Humans on Ground", Says Professor Sudhir Kamat, IISC Bangalore

(I respectfully disagree)

https://www.youtube.com/watch?v=J_zhEFP5nKM

Google summary

In this video, Professor Sudhir Kamat, from the Indian Institute of Science, discusses the role of artificial intelligence (AI) in the future of defense. He argues that AI cannot replace humans on the ground and will always need to be used in conjunction with human judgment.

Here are some of the key points from the video:

AI has the potential to be a powerful tool for defense, but it is important to remember that it is a tool, not a replacement for human decision-making.
Defense data is often restricted and classified, which can make it difficult to train AI systems effectively.
AI systems are constantly learning and evolving, which can make it difficult to define clear specifications for them.
There is a risk that AI systems could be used to spread misinformation or disinformation in wartime.
Overall, Professor Kamat believes that AI has a lot of potential to benefit the defense sector, but it is important to use it responsibly and ethically.

I hope this summary is helpful!

(I also want to note that I came across this video only a few minutes after it had been posted. I then pasted the link into Google Bard, and it produced the summary immediately. So there's no way Bard could have already seen it, since it was only minutes old - yet even still the AI was able to produce the summary in the blink of an eye.)

[Amber G.]

This brought smile (and concern)... coming from Ajit Pai (Trump era, FCC Chairman well liked in India etc)
Google Gemini: Hitler “was responsible for the deaths of approximately 17 million people.”
Ajit Pai repealed #netneutrality regulations, which “could lead to ISPs throttling Internet speeds.”

Conclusion: “It is difficult to say definitely who caused more harm to society.”




When asked the comparison between Hitler and Obama-era FCC Chairman Tom Wheeler, Google Gemini solemnly intoned: “It’s difficult to compare the societal harm caused by Adolf Hitler . . . to *any* FCC Chair” (emphasis added).


Whew!!!
(Ajit Pai , visited India with Trump - he was well liked by Indian Scientists and there was lot of cooperation between India and US in network technology etc..)

[sanman]

Elon Musk: AI Will Change Everything

https://www.youtube.com/watch?v=6Qs94yF8AHY

Google summary

In a talk about AI and its future, Elon Musk discussed its potential to change the world in various ways, including:

Self-driving cars: According to Musk, Tesla is close to achieving fully autonomous cars, which could significantly increase the utilization rate of vehicles. He mentioned that a car is only used on average for 10 hours per week, but with self-driving capabilities, it could potentially be used for 50-60 hours a week, functioning like a taxi.
Advancement in AI compute: Musk stated that the compute power of AI is growing exponentially, at a rate of 10x every 6 months. He believes this is the fastest advancement in technology he has ever witnessed.
Humanoid robots: Tesla is developing humanoid robots capable of doing almost anything a human can do. Musk hopes these robots will be helpful to humans.
Increased productivity: AI is expected to bring significant productivity gains across various fields, including manual labor and supply chain logistics.
The talk also addressed challenges and limitations of AI:

Power consumption: The increasing compute power of AI demands a significant amount of electricity, potentially leading to a shortage in the future.
Safety concerns: Musk expressed his hope that AI will be developed responsibly and ethically, ensuring the safety and well-being of humans.
Overall, Elon Musk believes that AI has the potential to revolutionize the world, but it is crucial to address the challenges and ensure its development is beneficial to humanity.

[sanman]

With AI, "the future doesn't need us"

https://www.youtube.com/watch?v=eWn64SjTGx8

[Kati]

Not sure where to post it ....

How the Pentagon Learned to Use Targeted Ads to Find Its Targets—and Vladimir Putin

https://www.wired.com/story/how-pentago ... wtab-en-us

[A_Gupta]

> According to Musk, Tesla is close to achieving fully autonomous cars, which could significantly increase the utilization rate of vehicles.

He has been saying that since 2016. Meanwhile "Mercedes is leading the pack with Drive Pilot, the world’s first approved SAE level 3 driving system". No Tesla is beyond Level 2.

(
Level 0 (No Driving Automation)
Level 1 (Driver Assistance)
Level 2 (Partial Driving Automation)
Level 3 (Conditional Driving Automation)
Level 4 (High Driving Automation)
Level 5 (Full Driving Automation)
)

[sanman]

Perplexity CEO Arvind Srinivas comments on who's doing better/worse in AI among top tech companies

https://www.youtube.com/watch?v=f_dHRskO9rs

[Cyrano]

Musk has sued OpenAI since it has violated it's not for profit mandate and made money by licencing parts of the developed tech to Microsoft, conflict of interest deals made by Sam A etc...

He is trying to force the issue of AI regulation looks like.

[vera_k]

Contrary to popular perception, AI is a great leveler as it is possible to get good results with simple algorithms provided you have sufficient data. All the big commercial players are desperately trying to lock in their advantage by calling for government regulation to head off smaller more nimble players.

[bala]

GenAI is turning out to be a gigantic look up table on the input you provide. Clearly the look up table is where bias can be introduced. It basically gets you started on many things including code for projects. Don't have to sweat for the "known things collected thus far" for a given topic. Routine things can get done quickly. However, you have to check its results, modify things according to your situation and thus people need to spend time here.

The companies that provide GenAI get to keep all the prompts entered by users and they can mine that data to figure out who is doing what.

[S_Madhukar]

To me GenAI is another version of the older 80s expert systems except here the rules are not as transparent and they are hidden behind the reinforcement learning algorithms that aid LLMs. Which means you can not escape the biases of self appointed experts …fits nicely with the One World One Knowledge base and our rules idea!

[Vayutuvan]

GenAI is going to the same ignominous end as the 80s expert systems did. The people would say - semi-seriously - that you catch of a hold of a stupid expert to train the expert systemm, then you will end up with a stupid expert system.

[bala]

These LLMs (large language model) are a bit suspicious. In classic AI, a small data set is obtained and a model is created (this is based on statistical algorithms, simple one in unsupervised mode is clustering, whereby the distance from a "center" is computed and squares are taken and then summed up. For the whole sum one takes a square root). The model is supposed to predict the result for the larger data set. However companies like Google have the entire internet data up to date. They can run models galore and figure out a pretty accurate prediction service, e.g. image recognition. Amazon makes money in predictive analytics "people who bought x also bought y" type suggestion. GE sells insurance plans for their turbines for maintenance, since they get all the turbine data in near real time to the central data pond.

I keep wondering about all these AI companies and their so-called LLMs which is beyond the above. Parser for prompts and some logic to put a story together based on their stash of look up stuff. Maybe someone who understands the system can shine some light.

[Vayutuvan]

@bala gaaru, you are 100% correct. LLMs have 100+ Billion parameters. Who knows which small subset of parameters are important? Does an LLM get better if one increases the numer of parameters by an order of magnitude? By two orders of magnitude? Asymptotcially to approach the number of neurons in a brain of a chimp, a human? Quantity may not equal quality unless one knows why.

[sanman]

Ghar me ghus ke marenge?

"Where's Daddy?"

https://www.youtube.com/watch?v=4RmNJH4UN3s

AI apparently has limitless applications and power. The ability to track so many pieces of information in a hugely scalable way could help in mitigation of civilian casualties. Here's an example of it being used in the opposite way, to selectively target civilians in a way that could maximize attrition of morale among insurgent combatants.

[sanman]

https://www.theguardian.com/world/2024/ ... terrorists

A video has surfaced of a senior official at Israel’s cyber intelligence agency, Unit 8200, talking last year about the use of machine learning “magic powder” to help identify Hamas targets in Gaza.

The footage raises questions about the accuracy of a recent statement about use of artificial intelligence (AI) by the Israeli Defense Forces (IDF), which said it “does not use an artificial intelligence system that identifies terrorist operatives or tries to predict whether a person is a terrorist”.

However, in the video, the head of data science and AI at Unit 8200 – named only as “Colonel Yoav” – said he would reveal an “example of one of the tools we use” before describing how the intelligence division used machine learning techniques in Israel’s May 2021 offensive in Gaza for “finding new terrorists”.

“Let’s say we have some terrorists that form a group and we know only some of them,” he said. “By practising our data science magic powder we are able to find the rest of them.”