A new bill in California aiming to regulate large frontier AI models has sparked significant resistance from various tech industry stakeholders, including startup founders, investors, AI researchers, and organizations advocating for open-source software. The bill, SB 1047, was introduced by California State Senator Scott Weiner.

 

Senator Weiner asserts that the bill mandates developers of large and powerful AI systems to adhere to common-sense safety standards. However, critics of the legislation argue that it would stifle innovation and jeopardize the entire AI industry.

 

In May, the California legislature passed the controversial bill, which is currently progressing through various committees. Following a final vote in August, the bill could be presented to Governor Gavin Newsom for signing into law. If enacted, SB 1047 would become the first major law in the United States regulating AI in a state that hosts many major tech companies.

 

Provisions of the Bill

 

Known as the 'Safe and Secure Innovation for Frontier Artificial Intelligence Systems Act,' SB 1047 aims to hold top AI companies like Meta, OpenAI, Anthropic, and Mistral accountable for the potential catastrophic dangers associated with rapidly advancing technology.

 

The bill specifically targets entities deploying large frontier AI models, defining "large" as those AI systems trained using computing power of 10^26 floating operations per second (FLOPS) with the training process costing more than $100 million. AI models fine-tuned using computing power greater than three times 10^25 FLOPS also fall under the bill's purview.

 

"If not properly subject to human controls, future development in artificial intelligence may also have the potential to be used to create novel threats to public safety and security, including by enabling the creation and the proliferation of weapons of mass destruction, such as biological, chemical, and nuclear weapons, as well as weapons with cyber-offensive capabilities," the bill states.

 

Liability and Compliance

 

According to the latest draft, developers behind large frontier AI models can be held liable for "critical harms," including using AI to create chemical or nuclear weapons and launching cyberattacks on critical infrastructure. It also covers human crimes committed by AI models with limited human oversight resulting in death, bodily injury, and property damage.

 

However, developers cannot be held responsible if AI-generated output leading to death or injury is information available elsewhere. The bill also mandates that AI models have an inbuilt kill switch for emergencies and prohibits the launch of large frontier AI models posing a reasonable risk of causing or enabling critical harm.

 

To ensure compliance, AI models must undergo independent audits by third-party auditors. Developers violating the bill's provisions could face legal action by California’s attorney general and must adhere to safety standards recommended by a new AI certifying body, the 'Frontier Model Division,' envisioned by the California government.

 

Controversy and Criticism

 

The draft legislation echoes concerns voiced by AI critics, including notable figures like Geoffrey Hinton and Yoshua Bengio, who believe AI could pose existential threats to humanity and thus requires regulation. The bill is also supported by the Centre for AI Safety, which published an open letter likening AI risks to those posed by nuclear wars or pandemics.

 

Despite this support, the bill has faced heavy criticism from many quarters. A major argument against it is that it could effectively eliminate open-source AI models. Open-source models ensure transparency and improved security as their inner workings can be freely accessed or modified by anyone. The proposed California bill, however, might discourage companies like Meta from making their AI models open source due to potential liability for misuse by other developers.

 

Conclusion

 

As California moves closer to potentially enacting SB 1047, the debate over regulating large frontier AI models continues to intensify. Proponents argue that the bill is necessary to safeguard against catastrophic harms posed by advanced AI technologies, while opponents fear it could stifle innovation and the open-source movement that has driven much of AI's progress.

 

The final outcome of this legislative effort will not only impact the tech industry in California but could also set a precedent for how AI is regulated across the United States and possibly the world. As the tech community grapples with these challenges, the balance between innovation and safety remains at the forefront of this critical discussion.

New Delhi - Major telecom operators Jio, Airtel, and Vi have announced significant hikes in their mobile data tariffs, affecting millions of consumers across India. The new plans from Jio and Airtel, which are up to 25% more expensive, take effect today. Meanwhile, Vi (Vodafone) will implement its tariff increase starting July 4.

 

 

Jio’s Revised Prepaid Plans: Jio has rolled out four new plans with 84 days of validity. The most affordable plan is now priced at Rs 479, up from Rs 395, offering 6GB of 4G data without 5G benefits. The Rs 799 plan, increased from Rs 666, provides 1.5GB of 4G data per day, also without 5G access.

 

The Rs 859 plan, up from Rs 719, and the Rs 1199 plan, up from Rs 999, offer 2GB and 3GB of 4G data per day, respectively, along with unlimited 5G access for the entire validity period.

 

Airtel’s New Prepaid Plans: Airtel has introduced three new plans with 84 days of validity. The most budget-friendly plan now costs Rs 509, up from Rs 455, providing 6GB of 4G data and unlimited calling, but no 5G benefits. The Rs 859 plan, previously Rs 719, offers 1.5GB of 4G data per day with unlimited calling, also lacking 5G access.

 

The Rs 979 plan, up from Rs 839, includes 2GB of 4G data per day, unlimited calling, and 5G network access, catering to heavy data users.

 

Vi’s Updated Prepaid Plans: Starting July 4, Vi will offer three new plans with 84 days of validity. The entry-level plan is now priced at Rs 509, up from Rs 459, offering 6GB of 4G data and unlimited calling. The Rs 859 plan, increased from Rs 719, provides 1.5GB of 4G data per day.

 

The top-tier plan, now costing Rs 979, up from Rs 839, offers 2GB of 4G data per day.

 

Comparative Analysis: Jio's most affordable plan starts at Rs 479, making it the cheapest option for limited data users. Both Jio and Airtel’s top-tier plans, offering unlimited 5G data, are priced similarly, appealing to heavy data consumers.

 

As the telecom giants grapple with rising costs and slowing subscriber growth, these tariff hikes are seen as a necessary move to sustain their operations. However, for the consumers, this means higher expenses for mobile data and connectivity.

The Rollout of Meta’s AI Chatbot

 

Last week, like many others, I noticed a blue ring icon appear at the top right corner above the search bar on WhatsApp. I had been anticipating the rollout of Meta’s new AI chatbot but was not particularly excited. In fact, Meta had been testing its AI chatbot for months in India, but now Meta AI is rolling out to the company’s main apps, including Facebook, Instagram, Messenger, and WhatsApp.

 

Honestly, my interest in AI chatbots peaked a long time ago. I have been occasionally using them for specific things, but they haven’t grown on me, nor have I become addicted to them. I still Google all my search queries, spend a lot of time on Reddit for Bollywood stuff, and get news from websites. That leaves less scope for a user like myself to make an effort to download an app or change my habits to log into a new service on a browser to use an AI chatbot.

 

Meta’s AI assistant came as a surprise when I had almost forgotten about the existence of these chatbots. Completely free to use, Meta AI resides inside WhatsApp, the app I use most in everyday life. And that is exactly why over the past week I have used this chatbot more than others over the past two years. This in-app integration could finally help familiarize many users with GenAI exactly because of this aspect. Here are the things I like and didn’t like about Meta’s AI chatbot.

 

 

The Fun of AI-Generated Images

 

Images created using AI have been a controversial topic, and there is definitely a need for standards regarding content provenance and ownership. However, in the context of sending AI-generated images to friends and family using a chat app like WhatsApp, this one seems harmless. Meta’s AI and its ability to generate images by typing a “description” make emojis so much fun. It’s safe too—the “Imagine with AI” label ensures the image is created using AI. For example, I typed, “imagine a photograph of a cat having fun in a pool in the summer.” In seconds, Meta AI created a convincing image. I felt Meta AI was faster in creating AI images than other image generators, which can take up to a minute to create an AI image. Sure, there are plenty of AI image generators, but having a feature built right into a chat app makes a lot of sense. Clearly, Meta is giving a large user base a user-friendly way to experiment with AI image generation without the risk of copying another artist’s style. The more users generate AI images and send them in messages, the more they become familiar with GenAI. To me, this may become Meta’s gateway to massify GenAI.

 

Finding Recipes Made Easy

 

I am not into food blogging, nor do I enjoy watching cooking videos online. However, it often happens that I hear about a specialty dish I had eaten in a restaurant and, out of curiosity, want to know the recipe. I remember watching a viral video of Deepika Padukone where she mentioned Ema Datshi, Bhutan’s national dish. It was a simple and easy-to-make dish at home. So, I asked Meta’s AI assistant to find a recipe for Ema Datshi. Meta pulled up the exact recipe I had seen in videos on social media. The good thing about Meta AI is that it shows the ingredients and instructions in an easy-to-digest format, without unnecessary information. I asked the AI assistant for a recipe that could be prepared at home with minimal ingredients and in less than 10 minutes. Meta AI responded with a recipe for Ema Datshi, which is super easy to prepare. But at this point, I wanted more options, so I typed, “ask for more suggestions.” Meta AI then came up with five recipes that can be made in less than 10 minutes. It was a major time saver, as I didn’t have to scroll through pages of food blogs or find videos on YouTube.

 

Shopping for Tech with Meta AI

 

When it comes to shopping for a tech product, I do not trust unreliable sources, especially AI chatbots, which have irked me in the past due to their limited knowledge and inability to search for up-to-date information. However, since Meta’s AI assistant uses Google Search and Bing to fetch results, I was willing to give it a try and see how it fares in shopping.

 

I told Meta AI that I currently own an iPhone 13 mini and am looking for a new phone under Rs 50,000. Meta produced a list of smartphones that can be bought under Rs 50,000. The recommendations were good and included a mix of both the newest phones and older smartphones. I was not expecting that, honestly. I quizzed Meta AI with many questions, and the AI assistant gave me the precise answers I was looking for. For instance, I asked Meta, “Should I switch to Android considering I use iMessage on my iPhone?” and Meta responded with this:

 

“If you’re deeply invested in the Apple ecosystem and rely heavily on iMessage, switching to Android might not be the most seamless experience. iMessage is exclusive to Apple devices, so you won’t be able to use it on an Android phone.”

 

I found the answer satisfactory. For context, Meta AI also recommended chat apps like WhatsApp and Telegram that are cross-platform and work on both iOS and Android, but I never felt it was forcing me to use only WhatsApp or suggesting it was better than iMessage.

 

Summarizing Articles

 

A lot of the time, if I am late to an event or need to catch a flight, I prefer to read short summaries rather than full news articles. It’s more convenient, even though sometimes depth is missing, but you still get the relevant information. I asked Meta to summarize a news article published on indianexpress.com titled “Indians spending more on foreign travel.” I cut and pasted the URL into my prompt, and Meta was able to summarize the article. The summary was not bare bones; it included all the necessary context, such as why Indians are traveling more, the reasons behind it, and the supporting numbers.

 

However, I found that Meta didn’t produce precise summaries of articles when it was an interview with many flowing quotes. I asked Meta to summarize an article I wrote during Apple’s Worldwide Developers Conference last month, which was based on a roundtable with the media and top executives. Instead of summarizing my article, Meta came up with a generic summary detailing Apple Intelligence features, far from what the article was about. To my surprise, Meta relied on other publications’ articles to pull results for the summary. The problematic part was that those articles lacked original reporting and did not even cover the discussion Apple’s executives had with the media, talking about all things Apple Intelligence. The only common thing between them and my article was the keyword “WWDC 2024.”

 

A Search Engine Alternative

 

For me, though, perhaps the biggest test for Meta AI was whether it can come close to Google in search queries. However, after spending some time with Meta’s AI assistant, I would say it’s definitely not replacing Google anytime soon but does basic search tasks well. For instance, I asked Meta to look up the nearest Bata store. It then asked me to share my current location or the city where I am, and when I did, it provided information about the nearest Bata stores around the location I specified. The results were correct, and the locations of the stores were in and around my home. Because I am familiar with my location, I can trust the results Meta AI came up with. However, I am not sure whether I can fully trust Meta if I am in a new city. I would still rely on Google. Another reason to go with Google search is that Meta does not show the location on a map, which is usually the first thing I do to find out how far the store is from my hotel.

 

Next, I asked Meta to find the best Chinese restaurants nearby. In response to my request, the chatbot replied it was already aware of my location but needed to know the preference I had in Chinese food.

 

“To recommend the best Chinese restaurants nearby, could you please tell me if you’re looking for a specific type of Chinese cuisine (like Szechuan or Cantonese), or if you have a preferred price range or dining experience (like delivery, takeaway, or dine-in)?”

 

Meta’s AI seemed chatty and it kept learning about my behavior to come up with precise results.

 

When I told Meta I was looking for a dine-in option, it recommended a few choices. However, except for one, none of them were great options. So I switched back to Google and used the web search I am familiar with to find better options. The point is that Google’s search engine is better for these types of search tasks because I can expand the search and keep refining it until I find exactly what I’m looking for. I can search for Chinese restaurants on Zomato or decide to opt for home delivery instead of dining in.

Smartphone cameras have seen remarkable advancements in the past decade, evolving from a modest 0.1 MP resolution to over 200 MP. Some high-end models now even feature 1-inch type sensors, offering image and video quality comparable to professional cameras.

 

Despite these improvements, smartphone cameras remain delicate. Simple mistakes can cause permanent damage. Here are five things to avoid to protect your smartphone camera system:

 

1. Shooting a Laser Light Show: Concerts often use high-intensity laser lights. Recording these with your smartphone can permanently damage the camera sensor due to their high energy density, affecting both the lens and the sensor.

 

2. Mounting on a Bike: Attaching your smartphone to a bike or scooter can cause permanent damage to the camera due to vibrations from the moving vehicle. Use a special mounting kit designed to absorb vibrations to prevent this.

 

3. Using a Smartphone Underwater: Despite some brands promoting underwater shooting capabilities, this can still permanently damage your phone, even with an IP rating. Extended underwater use heats up the phone, and condensation can lead to water entering the system, damaging the camera.

 

4. Shooting in Extreme Temperatures: Extreme temperatures, whether too hot or too cold, can damage the camera. Directly shooting the sun for extended periods, especially during solar eclipses, is particularly harmful. However, shooting the moon is generally safe, and some phones have dedicated moon modes.

 

5. Using Camera Lens Protectors: Poorly installed or low-quality lens protectors can scratch the camera lenses. Minor gaps between the lens and protector can trap dust particles, potentially causing damage.

 

Additionally, dropping your smartphone can harm camera features like stabilization and focusing capabilities.