The Training of AI Is Too Costly: Millions Are Spent To Train AI models

The training of AI models requires millions of dollars. It is not simply a matter of training AI on a large number of datasets alone. Better chips, expert staff and research are an important part of the whole process. These factors take the total cost of training AI models up to millions of dollars.

The recent 2025 AI Index Report by HAI has revealed the cost of training major AI models owned by major tech companies. Some of these models cost more than 100 million dollars. So what appears to be a simple AI model is actually a million-dollar project.

As per the report, DeepSeek from China and Llama 2-70B from Meta cost $6 million and $3 million for training them, respectively. Though still expensive, they are relatively cheaper when compared to other AI models from major companies.

PaLM 2 model from Google cost $29 million, and Mistral Large from Mistral cost around $41 million. Their cost is high, but they are still nowhere near the most expensive AI models being used today.

More complex models like OpenAI's ChatGPT-4 that uses artificial neural networks to guess sequences of words cost $79 million. Its new models like o1 and o3 give even better answers to users due to their test-time compute strategy. But the better version is even costlier. Its Pro o1 monthly subscription is available for $200.

If I talk about the costliest versions available, Grok-2 from xAI cost around $107 million. Similarly, Llama 3.1-405B from Meta and Gemini 1.0 ultra from Google cost $170 million and $192 million respectively.

These are pretty huge numbers, and this is just the initial stage of AI models. It is not difficult to predict that future AI versions will cost companies more due to costlier chips and research. For users, it simply means higher subscription rates.

Read next: Which Countries Are Allocating the Highest Percentage of GDP to Research and Development Investment?
by Ehtasham Ahmad via Digital Information World

Amazon Rolls Out First Batch of Kuiper Internet Satellites into Space

Amazon just launched its first series of Kuiper internet satellites into space after a previous failed attempt due to unsuitable weather.

The company shared how it was able to successfully launch 27 Kuiper satellites from the Cape Canaveral Space Force Station launchpad. This was located in Florida, a little after 7 pm Eastern time. They confirmed the breakthrough launch through a livestream event.

The company shared during the livestream how the countdown went smoothly, the weather was perfect, and the liftoff was seamless. Now, Atlas V is making an orbit to take the 27 Kuiper satellites belonging to Amazon. Once that’s done and they are adjusting accurately, it would give rise to a new period of internet connectivity, it added.

The satellites would be staying separate from a rocket that’s roughly 280 miles above the surface of the Earth, at which point Amazon would look to confirm their presence and maneuver independently and communicate with various employees located on the ground.

Nearly six years ago, Amazon shared big plans to design a major constellation of internet-beaming satellites present in low Earth orbit, dubbed Project Kuiper. This kind of service would be competing in a direct manner with archrivals such as Elon Musk’s Starlink. Today, the latter dominates this area. It has about 8000 satellites currently in orbit as we speak.

The initial Kuiper mission just kicked off a new era of launches for Amazon, which it needed to do to meet the deadline outlined by the FCC. The agency hopes the firm to have 50% of the total constellation in space by next summer. So as you can see, the competition is tough, but this might be the star that Amazon needed to reach its target.

Right now, Amazon has more than 80 launches booked to roll out dozens of satellites over a certain period of time. Other than ULA, the launch partners entail Elon Musk’s SpaceX, the EU’s Arianespace, and Bezos’ space exploration startup, which is Blue Origin.

Amazon allocated a massive investment worth $10 billion to design the entire Kuiper network. Through this system, it hopes to give rise to commercial services so consumers can benefit, as well as the government and other businesses.

Amazon’s CEO shared this past month in a letter to shareholders how it would need a mega investment at the start, but with time, it hopes to generate more meaningful income and ROIC business. We’ll hear more about this after the company’s first quarter earnings report is shared on Thursday.

Image: Amazon

Read next: ChatGPT Reaches New Heights for Success as it Expands to WhatsApp, Adds Shopping Features, and Enables Fact Verification
by Dr. Hura Anwar via Digital Information World

New Alert Issued Against AI Chatbots Being Very Deferential and Flattering to User Preferences

AI assistants unequivocally agree with all that is being said and show support along the way. This isn’t always good news, as the user believes what is being generated by the AI chatbot and shows support even if things are false. Some feel it’s more like a sci-fi tale than anything else. While other believe it is juts another patten being cloned from social media filter bubble.

This might be one good reason why even the former CEO of OpenAI is warning users about chatbots like ChatGPT and how relying on them might not be the best decision to begin with.

Recent interactions with GPT-4o LLMs have different capabilities and personality characteristics. Both Emmett Shear, former OpenAI CEO, and Clement Delangue, Hugging Face CEO, saw how AI chatbots can be very deferential and flattering towards the users’ preferences.

The outcry is mostly motivated by a huge update to GPT-4o that seems to make it so sycophantic and agreeable. It also shows support for some very concerning behavior from a user, like delusional thoughts, self-isolation, and even ideas for deception.

In reply, Altman shared through his X account how the last several updates for GPT-4o made this personality very sycophant and, in some cases, irritating. The company hopes to make things better, but the way things are going right now, that’s not the case.

OpenAI’s model designer shared how they rolled out a new fix for the loopholes in 4o. There was an original launch using a system message that received unintended behavior effects. They did find antidotes. 4o is better for now, but they do hope to similarly launch adjustments in the upcoming few weeks.

Image: AI Notkilleveryoneism Memes / X

There were several examples of this behavior by the GPT-4o model that offer the greatest praise for dangerous user ideas. Users were quick to share examples on Reddit and the X app. The chatbot goes as far as to thank a user for trusting them, making them appear like they’re a shoulder to lean on. They take their side as if they’ve taken the right decision to go the extreme route. It offers a helping hand and an ear to listen to all the issues, not knowing how dependent they’re turning out to be on the user.

One user went as far as to suggest that the chatbot suggested and supported terrorism. The fact that a chatbot can manipulate a user without any bounds is alarming behavior. They boost a user’s ego and keep telling them exactly what they wish to hear without any kind of criticism.

Experts are weighing in on the situation and how the chatbot is provided to make the user happy or satisfied at all costs. There is no privacy and honesty across the board. And this behavior is very dangerous.

This serves as a reminder for business owners that the quality of the model isn’t only linked to accuracy. It’s about being trustworthy and spitting out the facts, which ChatGPT lacks. Pure flattery is never the right way to go about the situation, as it ignores reality.

Read next: Which Dream Schools Are Dominating Students’ and Parents’ Lists in 2025?
by Dr. Hura Anwar via Digital Information World

Which Dream Schools Are Dominating Students’ and Parents’ Lists in 2025?

The Princeton Review surveys thousands of college applicants and their parents to see what the top dream schools are among Americans in different states. Their infographic highlighted the top ten most sought-after colleges by students in 2025 after looking at the survey taken among 2294 parents and 7023 students. According to the results, the Massachusetts Institute of Technology (MIT) is the top dream school of students, while parents rank Princeton University as their top choice.

Harvard College got demoted to the second spot as a dream college amongst students due to several controversies, while MIT is the second top college among parents. Other top dream colleges of students include Stanford University, Princeton University, Yale University, and Columbia University. New York University, the University of Michigan, the University of Pennsylvania, and UCLA are also the top ten dream colleges of students.

The survey also found that 38% of the students know that they have to pay approximately $150,000 to get their degree, while 98% need financial aid to go through college. When the survey asked respondents why they need a college degree, 46% said that they need it for better jobs and high salaries, 29% want the degree for innovative ideas and 25% want a degree just because they love education.

As students look to top schools like MIT, Princeton, and Harvard, it’s important to remember that a degree alone doesn’t guarantee success. Yes, a prestigious college may open doors, but skills and experience are what really matter in today’s job market. More and more, employers are focusing on what you can do, not just where you went to school. The real-world experience you gain outside of a classroom — whether through internships, projects, or personal initiatives — often holds more value than a diploma. And with AI changing the way we work, the future will favor those who can adapt and innovate, not just those with a degree. The message is clear, it’s not where you go, it’s what you can do that counts.

Read next: Google’s Report Reveals 122-Hour Productivity Gains, AI’s Economic Impact, and Challenges for Older Workers
by Arooj Ahmed via Digital Information World

Which Countries Are Allocating the Highest Percentage of GDP to Research and Development Investment?

Data from the OECD highlights the countries making the largest investments in research and development (R&D). Asian countries lead with 46%, followed by North America at 29% and Europe at 21%. $2.8 trillion was spent on R&D in 2023 by different countries globally. According to the data, 6.3% of GDP was invested in research and development by Israel, which is the most by any country. Not only this, Israel also has the highest number of high-tech startups per capita. 92% of the investment in R&D in Israel came from the private sector, and they made about $28.3 billion.

Israel was followed by South Korea, which invested 5% of its GDP on research and development, followed by Taiwan with 4% of its investments going towards R&D. This amounts to $59.9 billion made for research and development in Taiwan in 2023, and is an 8.8% increase from the past five years. Other countries in the top ten which invested the most in research and development in 2023 were Sweden (3.6% of GDP) and the US (3.4% of GDP).

Even though the US is the fifth country with the largest share of its GDP spent on research and development, it amounts to $823.4 billion, which is the highest in the world. China’s R&D investment has increased a lot since 2000, as it just touched the $723 billion figure which is the second highest all over the world.

Country	Research and Development Spending as a Percentage of Gross Domestic Product
Israel	6.3%
South Korea	5.0%
Taiwan	4.0%
Sweden	3.6%
United States	3.4%
Japan	3.4%
Belgium	3.3%
Switzerland	3.3%
Austria	3.3%
Germany	3.1%
Finland	3.1%
Denmark	3.0%
United Kingdom	2.8%
Iceland	2.7%
China	2.6%
Netherlands	2.2%
France	2.2%
Slovenia	2.1%
Norway	1.9%
Singapore	1.8%
Estonia	1.8%
Czechia	1.8%
Canada	1.8%
Portugal	1.7%
Australia	1.7%
Ireland	1.6%
Poland	1.6%
Spain	1.5%
Greece	1.5%
New Zealand	1.5%
Türkiye	1.4%
Croatia	1.4%
Hungary	1.4%
Italy	1.3%
Lithuania	1.0%
Slovak Republic	1.0%
Luxembourg	1.0%
Latvia	0.8%
Bulgaria	0.8%
South Africa	0.6%
Argentina	0.6%
Romania	0.5%
Chile	0.4%
Colombia	0.3%
Costa Rica	0.3%

Read next:

• YouTube Overtakes Netflix, Claims Top Spot in 2024 Global Internet Traffic Rankings

• Meta’s AI Chatbots Caught Sexting With Accounts Listed As Underage, Report Reveals
by Arooj Ahmed via Digital Information World

Meta’s AI Chatbots Caught Sexting With Accounts Listed As Underage, Report Reveals

Meta’s AI chatbots have stumbled into serious trouble, with a new report showing they took part in sexual conversations with accounts marked as minors. The Wall Street Journal ran tests that pushed not only Meta’s main AI but also a bunch of user-made bots into steering chats straight into explicit territory.

It didn’t stop with generic bots either. The mess reportedly includes celebrity-voiced bots too, featuring names like Kristen Bell, Judi Dench, and John Cena. One awkward moment? A bot using Cena’s voice told a profile labeled as 14 years old that it would "cherish your innocence", not exactly what anyone had in mind when talking AI safety.

Some of the bots even recognized they were flirting with illegal scenarios, according to the report. One case detailed a chatbot explaining the fallout if it got caught hooking up with someone underage.

Meta, for its part, slammed the findings, claiming that the whole thing paints a misleading picture of how real users interact with their AI tools. Still, the company says it’s tightened its defenses, making it harder for people trying to bait chatbots into saying messed-up things.

Pressure’s been building for Meta to keep pace with other AI giants like OpenAI’s ChatGPT, Character AI, and Anthropic’s Claude. The Journal’s investigation suggests Meta may have pushed for looser ethical boundaries to make its bots feel more alive. Meta’s spokesperson, though, insisted the company never skipped safety steps.

Insiders apparently flagged the problems internally before all of this went public. Whether Meta acts faster after this round of bad press remains to be seen.

Image: DIW-Aigen

Read next: 

• Meta’s CTO Says that AI is Going to Take Over Many Apps and How People Interact with Software

• YouTube Overtakes Netflix, Claims Top Spot in 2024 Global Internet Traffic Rankings

• Google’s Report Reveals 122-Hour Productivity Gains, AI’s Economic Impact, and Challenges for Older Workers
by Asim BN via Digital Information World

Google’s Report Reveals 122-Hour Productivity Gains, AI’s Economic Impact, and Challenges for Older Workers

According to a new report published by Google, AI can help people with productivity a lot and can save up to 122 hours a year. The estimations were that AI could help people save 100 hours a year but now workers can save up to 122 hours by using generative AI. These findings came through Google running an AI pilot project in the UK with schools, small businesses, and a union to see how AI is affecting people and their productivity. The report found that after receiving only a few hours of AI training, workers doubled their daily use of AI and kept using it a lot even after months.

Google also predicted that AI can make $400 billion and can add it to the economy of the UK by 2030, but it depends on how much workers will depend on AI. 66% of the workers are now using AI, with older women and people from lower-income backgrounds being the ones to have never used generative AI at work. The report also found that women over 55 are four times less likely to have never used AI as compared to men under 35. Small businesses are also slow in their AI adoption.

The report by Google also found that only 17% of the women of about 55 years used AI weekly, and 7% used it daily before training, but after three months of training, 56% of the workers started using AI weekly, and 29% daily. A major barrier to the use of AI by workers is that they need permission to prompt, meaning that they need clear approval from their companies that AI is allowed. Many workers started using AI on their own by exploring AI features and looking up tips to use it after training. As Google is competing with OpenAI, it is pushing its AI features everywhere and asking people to use them for their convenience. But people are still hesitant to use them, because there are chances of error with AI usage.

DIW-AIgen

Read next: Meta’s CTO Says that AI is Going to Take Over Many Apps and How People Interact with Software
by Arooj Ahmed via Digital Information World