Lica World, an AI startup in San Francisco, founded by Priyaa K and Purvanshi Mehta and backed by Replit CEO Amjad Masad, has unveiled an audio reel creation platform. Users can create audio reels for any form of written content including newsletters, research papers, emails, and documents. The platform thus makes podcasts in seconds!
Storytelling Platform
“Lights, Camera, Action,” or Lica, arose from the need to address common modes of office communication, such as powerpoint and other office products, that were built in the 90s and early 2000s and have not evolved since.
“We are building something where video storytelling, which is the most effective form of storytelling, can be democratised for people who don’t have access to the most powerful video editors or even have the knowledge to be able to use them effectively,” said CEO and co-founder, Priyaa, in an earlier exclusive interaction with AIM.
Priyaa believes that every time a story needs to be told, for example, a developer presenting a technical documentation, or a journalist writing a brief for their executive team, they are constrained within the paradigm of a tool.
With Lica, the founders look to enable people to become effective storytellers. The models will involve different levels of human intervention which will allow a person to create customised video depending on the occasion, and even fine tune at various stages to change background, voiceover sound, and many more as per one’s directorial style.
The startup is also backed by Replit VP of AI Michele Catasta, and Village Global, a venture capital firm chaired by Reid Hoffman and backed by Jeff Bezos, Bill Gates, Mark Zuckerberg and others.
Lica Testing
AIM demoed the platform with excerpts from a previous interview with Lica founders, and the results have been impressive. Lica was able to create a podcast of over 2-minute duration in a narrative style based on the type of voice selected (gender) with a background tone of your choice. It is possible that more options for user customisation can help fine-tune the product better.
Source: Lica Beta
The post Now You Can Create Podcasts in Seconds with Lica, Developed by Indian-Origin Founders appeared first on AIM.
A study conducted by Northwestern University revealed that prompt injection attacks on custom GPT models are highly effective, with a 97.2% success rate in extracting system prompts and a 100% success rate in leaking files.
Moreover, people frequently tend to share critical data with large language models (LLM). For instance, when Samsung lifted the ban on employees using ChatGPT, three data leaks were reported. In one of these, an employee prompted the entire source code of a faulty semiconductor database while trying to ask for help.
A report by LayerX, which analysed ChatGPT and other generative AI app usage for 10,000 employees, found that 6% of the employees pasted sensitive data into GenAI platforms such as OpenAI’s ChatGPT and Google’s Gemini, whereas 4% pasted such sensitive data on a weekly basis.
This put organisations on red alert, especially when data leaks are quite common through prompt injection attacks.
This makes it crucial for enterprises to have some sort of guardrails between the AI chatbot and the user, to identify the intentions of demanding censored data.
Firewall Matters (a lot)!
In a recent conversation with AIM, Ruchir Patwa from SydeLabs demonstrated how easy it is to prompt ChatGPT to get the desired information, be it the step-by-step instruction to make a bomb or rob a bank.
Patwa mentioned that even if you were to use an open-source LLM or a model built from scratch, you must have a guard between the prompt and the data. “In the AI era, intention matters more than data,” he said, further explaining how a crafted prompt injection goes beyond data.
This is where the idea of an AI firewall kicks in. An AI firewall can analyse a user’s prompt and prevent injection attacks and data exfiltration attempts, similar to vulnerabilities faced by traditional web and API applications.
This is one of the reasons why Cloudflare recently announced an AI firewall to safeguard organisations from prompt injections. Apart from Cloudflare, there are other companies like Nightfall AI providing similar services.
One would assume that adding a filter to the input prompt would stop the AI abuse, but it won’t. A skilled prompt engineer will always find new ways to manipulate the chatbot.
A Reddit user suggested that just filtering out prompt inputs will not work. “The API needs to be structured to prevent malicious attacks by design, instead of filtering them. You can’t filter out everything,” he added.
Businesses Already Implementing AI Firewalls
F5 has partnered with Prompt Security to deliver a firewall for AI applications on F5 Distributed Cloud Services, which can be easily deployed within F5 Distributed Cloud AppStack.
As mentioned earlier, Cloudflare has also developed an AI firewall to protect the traffic going through its network. However, it can also be deployed on models hosted on any other third-party infrastructure.
Darktrace, a cybersecurity company, has an in-house solution called Cyber AI Analyst to detect and prevent chatbot abuse.
While companies have started adopting AI firewalls, which will surely prevent attackers, filtering the output can also be an effective solution. This can be done by instructing the system to scan the output, rather than the input prompt itself.
The post How Can AI Firewall Safeguard Sensitive Data? appeared first on AIM.
IBM uses its annual Think conference as a platform to highlight movements in its business and present innovation, and this year was no exception. During this year’s Think event, held May 20-23 in Boston, the tech giant announced several updates to its watsonx platform, promising to make artificial intelligence more accessible, cost-effective and flexible for businesses.
We highlight the key announcements from IBM Think and detail the possible impact on IT pros in Australia in particular, with insights from Nick Flood, IBM’s managing director for Australia.
What are some key announcements from IBM Think?
IBM released a family of Granite models into open source and launched InstructLab in collaboration with Red Hat
Available under Apache 2.0 licenses on Hugging Face and GitHub, the open-source Granite models range from 3B to 34B parameters and come in both base and instruction-following model variants. Those variants are suitable for various tasks, including complex application modernization, code generation, fixing bugs, explaining and documenting code and maintaining repositories.
IBM unveiled a new range of watsonx assistants
IBM’s annual Global AI Adoption Index recently found that, while 42% of enterprise-scale companies (> 1,000 employees) surveyed have implemented AI in their businesses, another 40% of those companies that are exploring or experimenting with AI have yet to deploy their models.
To help those struggling to embrace AI, IBM announced the following upcoming updates and enhancements to its family of watsonx assistants:
watsonx Assistant for Z to transform how users interact with the system to quickly transfer knowledge and expertise. Availability is planned for June 2024.
An expansion of watsonx Code Assistant for Z Service with code explanation to help clients understand and document applications through natural language. Availability is planned for June 2024.
watsonx Code Assistant for Enterprise Java Applications. Availability is planned for October 2024.
IBM previewed new capabilities for AI-powered automation
At the event, IBM previewed a new generative AI-powered tool called IBM Concert, which will be generally available in June 2024. IBM claimed that Concert will serve as the “nerve centre” of an enterprise’s technology and operations. Powered by AI from watsonx, IBM Concert will provide generative AI-driven insights across clients’ portfolios of applications to identify, predict and suggest fixes for problems.
DOWNLOAD: AI quick glossary from TechRepublic Premium
In addition, IBM announced a wide range of activities with third parties, ranging from AWS and Microsoft to Adobe, Meta and SAP. In partnership with these companies, IBM is bringing third-party models onto watsonx, and offering IBM Consulting expertise for enterprise business transformation. This allows end-user customers adopt and scale AI solutions that are specific to their business needs.
What does this IBM news mean for Australia?
First and foremost, according to IBM’s Nick Flood, the company’s announcements will help Australians grapple with several headline issues in the economy, including the skills shortage and a lack of productivity.
“The number one issue on the minds of boards and elected officials in Australia is productivity, specifically Australia’s lack of it,” Flood said. He cited data from the Centre for Economic Development of Australia and the OECD, where Australia ranks as low as 61 out of 63 counties in some areas. “We’re going backwards,” Flood said, “and comparatively, we’re lagging behind comparable countries like the United Kingdom, United States.”
Flood believes that taking a leadership position in generative AI and quantum computing can help Australia leapfrog ahead in terms of productivity. “I’m full of optimism that, with great Aussie ingenuity and these emerging technologies, we can leapfrog ahead in terms of productivity. And everyone’s going to benefit from that.”
Flood also shared exciting work around how generative AI is helping in the Australian setting overcome critical skill shortages. “At our THINK conference, we launched two generative AI capabilities specifically for IBM mainframe technology,” he said. These include watsonx Assistant for Z, a generative AI-based chat agent that can generate contextualised run sheets or recommendations, and include watsonx Code Assistant for Z, which can take legacy mainframe architectures and code and rewrite them into more contemporary programming languages, all without human intervention.
Challenges to AI and IBM’s vision in Australia
Despite his optimism, Flood acknowledged the challenges that stem from AI adoption in Australia. “One of the other headlight topics on the minds of boards and CIOs, CEOs, elected officials is cyber risk,” he said.
Hyperscale cloud platforms are still a roadblock
He noted that Australian clients who want to step into generative AI are hesitating when they are required to make exclusive use of hyperscale cloud platforms. They desire a setup in which they could live in a hybrid setup where certain large language models would sit securely on Australian shores in their data centre.
Flood highlighted the importance of understanding these technologies and how they can be applied to work. “I think the number one challenge is really ensuring that there’s a consistency of understanding across the organisation and across all operations or disciplines, not just in IT, around what generative AI could do for the enterprise,” he said.
Social responsibility is key to AI
He also acknowledged that there were some social challenges that need to be kept in mind, particularly in multicultural societies such as Australia. Flood said it’s imperative that, as IT professionals build and roll out AI, it’s free of bias and aligned to the highest ethical standards.
“IBM is taking a lot of proactive steps to engage with government to ensure that as AI proliferates across the economy, there’s safeguards in place, and there’s discussion initiated to think about both the intended and the unintended consequences and how best to govern the detrimental impacts of the latter.”
SEE: 9 Innovative Use Cases of AI in Australian Businesses in 2024
IBM’s commitment to Australia includes supporting quantum computing
Finally, Flood made it clear that IBM is not a bystander in the Australian market; the company established operations in 1932, and now maintains around 3,600 staff in the country. Far from being a branch or sales office, IBM’s local team generates patents and supports customers on a technical level.
Currently, this means IBM is actively supporting Australia to achieve its quantum computing ambitions. “Earlier in the month, IBM and the University of Sydney won a US $10 million award from IAPRA. Sydney University researchers will work now with IBM researchers and use the IBM quantum capability delivered over the IBM cloud from Yorktown Heights in upstate New York to develop new mechanisms around quantum error suppression, which is a really key milestone that the world will need to overcome on the path to quantum utility,” Flood said.
“Also, earlier this month IBM was delighted to be part of a consortia with the University of Sydney that won a $18.3 million Australian grant from the Albanese government to develop the Australian Centre for Quantum Growth at the University of Sydney.
“We are passionate about the national interest,” Flood said. “We’re very proud of what we have achieved in this country.”
Stability AI, the startup behind the AI-powered art generator Stable Diffusion, has released an open AI model for generating sounds and songs that it claims was trained exclusively on royalty-free recordings.
Called Stable Audio Open, the generative model takes a text description (e.g. “Rock beat played in a treated studio, session drumming on an acoustic kit”) and outputs a recording up to 47 seconds in length. The model was trained using around 486,000 samples from free music libraries FreeSound and the Free Music Archive.
Stability AI says that the model can be used to create drum beats, instrument riffs, ambient noises and “production elements” for videos, films and TV shows as well as to “edit” existing songs or apply the style of one song (e.g. smooth jazz) to another.
“A key benefit of this open source release is that users can fine-tune the model on their own custom audio data,” Stability AI wrote in a post on its corporate blog. “For example, a drummer could fine-tune on samples of their own drum recordings to generate new beats.”
Stable Audio Open has its limitations, however. It can’t produce full songs, melodies or vocals — at least not good ones. Stability AI says that it’s not optimized for this, and suggests that users looking for those capabilities opt for the company’s premium Stable Audio service.
Stable Audio Open also can’t be used commercially; its terms of service prohibit it. And it doesn’t perform equally well across musical styles and cultures or with descriptions in languages other than English — biases Stability AI blames on the training data.
“The source of data is potentially lacking diversity and all cultures are not equally represented in the data set,” Stability AI writes in a description of the model. “The generated samples from the model will reflect the biases from the training data.”
Stability AI — which has long struggled to turn its flagging business around — became the subject of controversy recently after its VP of generative audio, Ed Newton-Rex, resigned over disagreement with the company’s stance that training generative AI models on copyrighted works constitutes “fair use.” Stable Audio Open would appear to be an attempt to turn that narrative around, while at the same time not-so-subtly advertising Stability AI’s paid products.
As music generators including Stability’s gain in popularity, copyright — and the ways in which some creators of generators might be abusing it — is becoming a central point of focus.
In May, Sony Music, which represents artists including Billy Joel, Doja Cat and Lil Nas X, sent a letter to 700 AI companies warning against “unauthorized use” of its content for training audio generators. And in March, the U.S.’ first law aimed at tamping down abuses of AI in music was signed into law in Tennessee.
We're just a week away from finally learning how Apple plans to add a dose of AI to its core products — and where it'll stack up compared to Google, OpenAI, and Microsoft, all of which have already hosted their spring developer conferences.
Also: 6 ways Apple can leapfrog OpenAI, Microsoft, and Google at WWDC 2024
This year's Worldwide Developers Conference, or WWDC, will take place starting Monday, June 10, and wrap up on June 14. The opening day is when the big keynote happens, with CEO Tim Cook and several executives taking the stage to announce the for-consumer updates. The days following are dedicated to developer workshops and private demo sessions.
Naturally, developers and members of the press will be in attendance at Apple Park in Cupertino throughout the week, while everyone else can catch a live stream of the opening keynote, either on Apple's website or YouTube channel.
What is expected at WWDC 2024?
WWDC is typically the event in which Apple takes the wraps off the next major versions of its assorted operating systems. That means we should anticipate demos of iOS 18, iPadOS 18, MacOS 15, WatchOS 11, tvOS 18, and VisionOS 2.0.
The event provides developers with access to experts, along with highlights of new tools and features that will help them create new and/or better apps for the Apple ecosystem.
Also: 10 things I'd like to see in VisionOS 2.0
"We're so excited to connect with developers from around the world for an extraordinary week of technology and community at WWDC24," Susan Prescott, Apple's VP of Worldwide Developer Relations, said in a news release. "WWDC is all about sharing new ideas and providing our amazing developers with innovative tools and resources to help them make something even more wonderful."
1. You'll be hearing 'AI' a lot
This year's WWDC promises something extra, namely a spotlight on Apple's endeavors into AI. With companies such as OpenAI, Microsoft, and Google already infusing their products with generative AI, Apple is clearly behind in the race. Even if consumers aren't longing for AI enhancements to all their usual apps and services, investors are anxiously waiting to see what the company can pull off in this new era of technology.
To catch up, Apple reportedly has been working on its own in-house AI tech to add to the next-generation iPhone and other products. On tap at WWDC might be AI-based assistance for services like Apple Music and a major and much-needed overhaul for Siri.
Also: 5 rumored iOS 18 features I'm most excited about — and AI is just the start
But Apple has also allegedly been seeking a partner for outside help, possibly teaming up with OpenAI to bring its chatbot expertise to iOS and Google to bring Gemini-powered AI features. Just a few months ago, the company purchased a Canadian startup firm called DarwinAI, which has designed ways to make AI systems smaller and more efficient.
More recently, rumors have suggested that some new AI features will include the ability to transcribe and summarize in Voice Memos and Notes, more intelligent and helpful searches in Safari, AI-generated emojis based on conversations in Messages, and more.
2. Don't forget the other acronym: RCS
To the surprise of many, except for the European Commission, Apple announced last year that iPhones would eventually support Rich Communication Services (RCS), a protocol already adopted by Android phones. Adding this technology should alleviate key pain points when messaging between the two operating systems, including the lack of typing indicators, disorientated group chats, and quality loss when sending media files.
Also: DOJ sues Apple: What it could mean for iPhone users and iOS developers
The decision to bring RCS to the iPhone came after mounting pressure from the European Union's Digital Markets Act (DMA), which stressed cross-platform compatibility. While a more recent statement from Google suggested that Apple would integrate RCS later this fall, highlighting the transition at WWDC could potentially help Apple's defense against the DOJ's antitrust lawsuit, filed in March. Regardless of when and how Apple chooses to announce the new feature, it'll be big news for both iOS and Android users.
Alongside iOS, expect AI feature upgrades across Apple's software portfolio, including the now two-year-old VisionOS. Considering the company's push to reposition the MacBook as the go-to AI PC, Apple will likely carry over some of the new Siri and AI functionalities for iOS introduced earlier in the event to MacOS 15. Likewise, iPadOS 18 is expected to receive an AI makeover that brings improved multitasking capabilities — possibly to Stage Manager — and a new eye-tracking accessibility feature.
As for VisionOS and Apple's constant pursuit of marketing its $3,500 Vision Pro headset, expect subtle, quality-of-life enhancements, including the ability to move apps around in the home screen, more first-party services, and a more flexible user experience in general.
At the keynote speech of Zoho Corporation’s user developer conference, Zoholics ‘24, that is currently ongoing in Austin, co-founder and CEO Sridhar Vembu commented on Salesforce’s recent disappointing earnings that led to a huge drop in its market capitalisation, alluding to the company’s weak planning strategy.
Showcasing quotes from Salesforce President and COO Brian Millham during the company’s recent earnings call, including ‘measured buying behaviour’ and ‘elongated deal cycles, deal compression, and high levels of budget scrutiny,’ Vembu described them as, “The words that erased $52 billion in market capitalisation in 24 hours.”
He even jokingly said, “An AI can regurgitate all those words now, about what a good earnings warning is like.”
Last week, Salesforce announced its disappointing earnings results which led to its stock falling by almost 20% the next day. It was the biggest single-day percentage drop since its stock fell 27.2% in July 2004. The company lost $52 billion in a day.
Talking about Zoho’s promising records, Vembu called out that they have zero debt on their balance sheet. “Our balance sheet is one of the cleanest in the world,” he said.
Profits and Firing
Vembu has often been critical about other companies’ approach when it comes to firing employees. Believing that even a 7% attrition rate at Zoho is terrible, Vembu was amazed at how one of the founders he interacted with was having no problem with a 25% rate in their company.
Vembu pointed out how a number of profitable companies last year resorted to laying off people which he considered to be stupid. “I don’t understand why they do it because they don’t understand what it does to employee morale long term.”
Vembu explains how companies cutting costs to survive makes sense, whereas, profitable companies that say they are not making enough profit or enough numbers and growth, hence, resorting to firing is incomprehensible.
“I won’t be able to make these statements as a public company CEO today, which tells you something about what our priorities have become in business. We want to escape that type of pressure, that’s why we have chosen to stay private,” said Vembu.
Interestingly, this is not the first time Vembu criticised Salesforce. Earlier this year, Vembu called out Salesforce’s hiring and firing sprees to be driven by short-term, quarterly financial goals.
Source: X
Scrutinising Zoho’s valuation, Sridhar Vembu remains nonchalant. “I pay absolutely no attention to what our valuation is. I don’t know and I don’t care,” he said.
The post Sridhar Vembu Takes a Dig at Rival Salesforce’s Market Cap Decline appeared first on AIM.
AI is rapidly evolving, and in order to stay relevant it’s important that you match the pace and keep yourself updated with the latest AI advancements.
To facilitate this, The Association of Data Scientists (ADaSci) offers a variety of AI courses designed to cater to different expertise levels, from mastering LangChain and building AI agents to understanding RAG and parameter-efficient fine-tuning.
Whether you’re a beginner in the GenAI field or a seasoned AI professional, these courses provide hands-on experience and detailed knowledge to keep you ahead in the game. ADaSci’s unique courses are not available anywhere else.
Discover the top 12 AI courses available on ADaSci and unlock new opportunities.
Generative AI Crash Course with Hands-on Implementations
This course will help you get an in-depth understanding of GenAI and its popular models. Participants will receive a detailed knowledge of GPT models, diffusion models, different NLP transformers and ChatGPT. The course will further provide you with a hands-on knowledge of implementing GenAI models in real-world applications.
This course caters to everyone, from beginners in GenAI looking to deepen their understanding and practical skills to professionals in AI and related fields seeking to update their knowledge with the latest advancements in GenAI.
Mastering LangChain: A Hands-on Workshop for Building Generative AI Applications
This LangChain workshop will help participants master GenAI for innovative applications across industries. You will learn to build and deploy custom AI agents, leveraging LangChain for transformative personalised solutions.
Participants should have a foundational understanding of AI and basic programming skills, preferably in Python.
Diving Deeper into Retrieval-Augmented Generation (RAG) with Vector Databases
This course will help you master the core principles of RAG and its advantages over pure generative models. Participants will delve into advanced AI techniques, unlocking the synergy between RAG and vector databases. You will also understand the tools and strategies for building, deploying, and optimising RAG systems.
Parameter-efficient Fine-tuning of Large Language Models
This workshop will help you understand Parameter-efficient fine-tuning (PEFT) techniques and their benefits for LLM adaptation. Participants will learn methods like LoRA, adapters, and prompt tuning to achieve remarkable results using less parameters.
You will also get hands-on experience building and evaluating your own PEFT model on provided datasets. With this course, you can master resource-efficient training strategies and deployment options for PEFT models.
Building Generative AI Applications with Amazon Bedrock
This hands-on course will provide you with a solid understanding of the Amazon Bedrock architecture, capabilities, and applications. It will help participants develop skills in building and deploying GenAI applications on Bedrock, allowing them to gain insights into real-world use cases, best practices, and the future potential of Bedrock.
Mastering Prompt Engineering for LLMs
With this course, participants will understand the fundamentals of prompt engineering and master the art of crafting, optimising, and customising prompts for various AI models.
It will help you explore various prompting concepts and techniques such as Zero-shot and Few-shot Prompting, Chain of Thought Prompting, Knowledge Generation Prompting, and more.
The LLMOps : Streamlining the GenAI & LLM Operations
This course can be beneficial in understanding the fundamentals of LLMOps and its role in GenAI-powered systems and NLP.
Participants will develop knowledge about the workings of LLMOps and explore its challenges such as model training, deployment, monitoring, and maintenance. They will also learn the design process of LLMOps and acquire practical skills in innovating within the LLMOps operations.
Autonomous AI Agents and AI Copilots
This course will teach you the foundational concepts behind building AI agents and delve into different ML techniques that make them smarter. It also examines the challenges of creating dependable AI agents and the ethical considerations that come with them.
Through this course, you’ll be able to analyse the potential benefits and limitations of autonomous AI agents and AI copilots in different application domains such as healthcare, finance, creative work etc.
You will also understand various techniques in autonomous AI agents and copilots such as BabyAGI, MetaGPT, and Semantic Kernels.
Advanced RAG with Pinecone
This course will take your text generation skills to the next level. It will help you master the utilisation of Pinecone for information retrieval in RAG.
You’ll learn about integrating knowledge bases and crafting powerful prompts, creating informative and creative text outputs.
Building Multi-Agent LLMs with AutoGen
With this course, you’ll learn how to build multi-agent LLMs and create collaborative AI systems using the AutoGen framework.
It will also help you unlock real-world applications, exploring how multi-agent LLMs can be applied in various domains for problem-solving.
Vector Search Techniques with Weaviate
This course will help you explore advanced vector search techniques using Weaviate, a vector search engine. You will learn about Weaviate’s architecture, features, and capabilities for vector-based search and semantic querying.
Participants will dive into hands-on exercises to master indexing, querying, and optimising vector search performance.
Generative AI Application Development with Azure
This course equips you with essential skills to develop, deploy, and monitor GenAI applications using Microsoft Azure. You will gain hands-on experience with Azure’s powerful AI services, enhance your technical expertise, and learn to develop scalable AI solutions.
The post Top 12 Generative AI Courses Available on ADaSci appeared first on AIM.
Hiya, folks, and welcome to TechCrunch’s inaugural AI newsletter. It’s truly a thrill to type those words — this one’s been long in the making, and we’re excited to finally share it with you.
With the launch of TC’s AI newsletter, we’re sunsetting This Week in AI, the semiregular column previously known as Perceptron. But you’ll find all the analysis we brought to This Week in AI and more, including a spotlight on noteworthy new AI models, right here.
This week in AI, trouble’s brewing — again — for OpenAI.
A group of former OpenAI employees spoke with The New York Times’ Kevin Roose about what they perceive as egregious safety failings within the organization. They — like others who’ve left OpenAI in recent months — claim that the company isn’t doing enough to prevent its AI systems from becoming potentially dangerous and accuse OpenAI of employing hardball tactics to attempt to prevent workers from sounding the alarm.
The group published an open letter on Tuesday calling for leading AI companies, including OpenAI, to establish greater transparency and more protections for whistleblowers. “So long as there is no effective government oversight of these corporations, current and former employees are among the few people who can hold them accountable to the public,” the letter reads.
Call me pessimistic, but I expect the ex-staffers’ calls will fall on deaf ears. It’s tough to imagine a scenario in which AI companies not only agree to “support a culture of open criticism,” as the undersigned recommend, but also opt not to enforce nondisparagement clauses or retaliate against current staff who choose to speak out.
Consider that OpenAI’s safety commission, which the company recently created in response to initial criticism of its safety practices, is staffed with all company insiders — including CEO Sam Altman. And consider that Altman, who at one point claimed to have no knowledge of OpenAI’s restrictive nondisparagement agreements, himselfsigned the incorporation documents establishing them.
Sure, things at OpenAI could turn around tomorrow — but I’m not holding my breath. And even if they did, it’d be tough to trust it.
News
AI apocalypse: OpenAI’s AI-powered chatbot platform, ChatGPT — along with Anthropic’s Claude and Google’s Gemini and Perplexity — all went down this morning at roughly the same time. All the services have since been restored, but the cause of their downtime remains unclear.
OpenAI exploring fusion: OpenAI is in talks with fusion startup Helion Energy about a deal in which the AI company would buy vast quantities of electricity from Helion to provide power for its data centers, according to the Wall Street Journal. Altman has a $375 million stake in Helion and sits on the company’s board of directors, but he reportedly has recused himself from the deal talks.
The cost of training data: TechCrunch takes a look at the pricey data licensing deals that are becoming commonplace in the AI industry — deals that threaten to make AI research untenable for smaller organizations and academic institutions.
Hateful music generators: Malicious actors are abusing AI-powered music generators to create homophobic, racist and propagandistic songs — and publishing guides instructing others how to do so as well.
Cash for Cohere: Reuters reports that Cohere, an enterprise-focused generative AI startup, has raised $450 million from Nvidia, Salesforce Ventures, Cisco and others in a new tranche that values Cohere at $5 billion. Sources familiar tell TechCrunch that Oracle and Thomvest Ventures — both returning investors — also participated in the round, which was left open.
Research paper of the week
In a research paper from 2023 titled “Let’s Verify Step by Step” that OpenAI recently highlighted on its official blog, scientists at OpenAI claimed to have fine-tuned the startup’s general-purpose generative AI model, GPT-4, to achieve better-than-expected performance in solving math problems. The approach could lead to generative models less prone to going off the rails, the co-authors of the paper say — but they point out several caveats.
In the paper, the co-authors detail how they trained reward models to detect hallucinations, or instances where GPT-4 got its facts and/or answers to math problems wrong. (Reward models are specialized models to evaluate the outputs of AI models, in this case math-related outputs from GPT-4.) The reward models “rewarded” GPT-4 each time it got a step of a math problem right, an approach the researchers refer to as “process supervision.”
The researchers say that process supervision improved GPT-4’s math problem accuracy compared to previous techniques of “rewarding” models — at least in their benchmark tests. They admit it’s not perfect, however; GPT-4 still got problem steps wrong. And it’s unclear how the form of process supervision the researchers explored might generalize beyond the math domain.
Model of the week
Forecasting the weather may not feel like a science (at least when you get rained on, like I just did), but that’s because it’s all about probabilities, not certainties. And what better way to calculate probabilities than a probabilistic model? We’ve already seen AI put to work on weather prediction at time scales from hours to centuries, and now Microsoft is getting in on the fun. The company’s new Aurora model moves the ball forward in this fast-evolving corner of the AI world, providing globe-level predictions at ~0.1° resolution (think on the order of 10 km square).
Image Credits: Microsoft
Trained on over a million hours of weather and climate simulations (not real weather? Hmm…) and fine-tuned on a number of desirable tasks, Aurora outperforms traditional numerical prediction systems by several orders of magnitude. More impressively, it beats Google DeepMind’s GraphCast at its own game (though Microsoft picked the field), providing more accurate guesses of weather conditions on the one- to five-day scale.
Companies like Google and Microsoft have a horse in the race, of course, both vying for your online attention by trying to offer the most personalized web and search experience. Accurate, efficient first-party weather forecasts are going to be an important part of that, at least until we stop going outside.
Grab bag
In a thought piece last month in Palladium, Avital Balwit, chief of staff at AI startup Anthropic, posits that the next three years might be the last she and many knowledge workers have to work thanks to generative AI’s rapid advancements. This should come as a comfort rather than a reason to fear, she says, because it could “[lead to] a world where people have their material needs met but also have no need to work.”
“A renowned AI researcher once told me that he is practicing for [this inflection point] by taking up activities that he is not particularly good at: jiu-jitsu, surfing, and so on, and savoring the doing even without excellence,” Balwit writes. “This is how we can prepare for our future where we will have to do things from joy rather than need, where we will no longer be the best at them, but will still have to choose how to fill our days.”
That’s certainly the glass-half-full view — but one I can’t say I share.
Should generative AI replace most knowledge workers within three years (which seems unrealistic to me given AI’s many unsolved technical problems), economic collapse could well ensue. Knowledge workers make up large portions of the workforce and tend to be high earners — and thus big spenders. They drive the wheels of capitalism forward.
Balwit makes references to universal basic income and other large-scale social safety net programs. But I don’t have a lot of faith that countries like the U.S., which can’t even manage basic federal-level AI legislation, will adopt universal basic income schemes anytime soon.
Groundwater is a vital water supply for humanity. It supplies 35% of the drinking water in the U.S. and almost half of all drinking water in the world. Unfortunately, over-extraction, industrial contamination, and pollution from agricultural activities have depleted the groundwater resources. To make things worse, traditional methods for groundwater assessments have proven inadequate for accurately monitoring and managing these resources.
A research team led by the San Diego Supercomputer Center (SDSC) at UC San Diego is using data science and AI to develop a more in-depth understanding of groundwater dynamics. This could enable the researchers to lay the foundation for advanced models to assess groundwater resources.
Principal Investigator (PI) Ilya Zaslavsky, director of the SDSC Spatial Information Systems Laboratory, is heading the SDSC team, along with Co-PI Christine Kirkpatrick, director of the SDSC Research Data Services Division, and Co-PI Ashley Atkins, SDSC chief of staff. The SDSC team is joined by researchers from Poland, Ukraine, Lithuania, Latvia, Estonia, and the U.S.
The two-year project, titled Groundwater Resilience Assessment through Integrated Data Exploration for Ukraine (GRANDE-U), is funded jointly by the U.S. National Science Foundation (NSF), Research Council of Lithuania (LMT), Estonian Research Council (ETAG), Latvian Council of Science (LCS), National Science Center of Poland (NCN), U.S. National Academy of Sciences and Office of Naval Research Global (DoD).
Assessing and managing groundwater is particularly challenging in transboundary regions, where aquifers span across political boundaries. The challenges of international coordination coupled with a few technical issues make it difficult for researchers to conduct reliable groundwater assessments.
The uneven sensor networks, disparate data collection methods, and incompatible hydrogeologic descriptions across the transboundary regions make it complex to model groundwater systems and manage the aquifers effectively.
According to Zaslavsky, one of the primary objectives of the research team is to “integrate hydrogeologic models with satellite and ground-based observations and deliver highly detailed and timely predictions of groundwater storage and flows across borders.”
The SDSC GRANDE-U team aims to use the power of AI and data science to overcome the limitations of traditional methods for groundwater observation. One of the new methods being used is remote sensing for groundwater assessment using aerial imagery to gather information about the regions’ subsurface features.
The remote sensing method is even more vital for regions like Ukraine, where there is a sharp rise in demand for aquifers for drinking water, but the population displacement in the eastern parts of the country has made it difficult for researchers to gather critical data to understand groundwater dynamics.
According to Oleksii Shevchenko, chief scientist at the Ukrainian Hydrometeorological Institute in Kyiv and leader of the Ukrainian research team, addressing the groundwater assessment challenges transcends the capabilities of any single nation or discipline. The project’s success is heavily reliant on collaboration and knowledge-sharing across borders.
“Beyond the technical challenges of groundwater hydrology, this project also delves into the economic, social, and political factors affecting groundwater storage and dynamics,” said Co-PI Atkins. “We aim to understand the role that perceptions play in water decision-making and by working with transboundary water resource experts on our international team, we strive to enhance the credibility and reliability of our models.”
via Shutterstock
AI and data science could play a key role in unlocking innovative solutions to addressing the threats of groundwater depletion. Researchers around the globe are using these technologies to gain a deeper understanding of groundwater dynamics.
The United States Geological Survey (USGS) has been developing AI models to predict groundwater responses to various stressors, such as climate change. NASA has been working on combining remote sensing and AI techniques for aquifer mapping and its recharge dynamics. AI is also being used in several other projects to save Earth’s precious resources.
The global research efforts demonstrate the potential of AI and data science in sustainable groundwater management. As the demand for groundwater continues to grow, advanced technologies will remain crucial for tackling the pressing challenges of water scarcity and resource management.
Related Items
Scientists Use GenAI to Uncover New Insights in Physics
Lightning AI Today Announces ‘Thunder’ to Speed Up AI Model Training
I'm sure you're familiar with the inexpensive Raspberry Pi single board computer. I have ten or so of them doing all sorts of specialized tasks around the workshop and Fab Lab. As of February, Raspberry Pi has sold 61 million of the devices over a 12 year period since launch.
These devices aren't just used the way I use them in hobbyist experiments or to control 3D printers and smart home features. Clusters of hundreds of Pis have been used to test supercomputer software. They're widely used for industrial automation control. They're embedded inside commercial products. Back in 2020, Raspberry Pis even powered ventilators keeping COVID-19 patients alive.
Also: Connect to your Raspberry Pi from a web browser, anywhere in the world — here's how
Mixing AI with the Raspberry Pi architecture is nothing new. But now, Raspberry Pi is shipping its own $70 AI Kit, based on hardware from edge chipmaker Hailo. Adrian Kingsley-Hughes discussed the tech in this kit in his article on the topic. He's intrigued by the camera and pose analysis capabilities, with an eye towards upgrading his security system.
Just before the launch, I had the opportunity to sit down with Eben Upton, CEO and co-founder of Raspberry Pi and Orr Danon, CEO and co-founder of Hailo to talk about the product, the launch, AI, and what it all means.
And with that, let's get started.
ZDNET: What inspired the partnership between Hailo and Raspberry Pi?
Eben Upton (Raspberry Pi): Raspberry Pi and Hailo are on very similar missions: to bring the power of general-purpose computing and of high-performance high-efficiency AI acceleration to a broad community of enthusiasts, educators, and industrial and embedded customers. So it seemed like a natural collaboration, and we've been working for the last six months to make it a reality.
ZDNET: How does the Raspberry Pi AI Kit differ from previous Raspberry Pi products?
Eben Upton (Raspberry Pi): Our customers have been building AI applications on our products almost since we launched Raspberry Pi 1 back in 2012. These either ran on our increasingly capable Arm CPU cores, or on older medium-performance accelerators, of which Google Coral is the most obvious example.
The AI Kit breaks new ground, both as our first Raspberry Pi-branded AI product, and by pushing performance up substantially in comparison with those legacy accelerators. We think it's going to enable a new wave of AI applications both in our hobbyist community and in the industrial and embedded space.
ZDNET: What are some key features of the Hailo-8L M.2 AI acceleration module?
Orr Danon (Hailo): The Hailo-8L provides 13 tera operations per second of compute power, enabling the Raspberry Pi 5 to run advanced AI modules, mainly vision-based applications for video analytics. This is done in a very low power consumption which does not require active cooling of the device, empowering Raspberry Pi users with state-of-the-art AI capabilities.
ZDNET: Can you share some examples of innovative projects that could be developed with the Raspberry Pi AI Kit?
Orr Danon (Hailo): There are endless options to embed video analytics and AI algorithms into Raspberry Pi applications, and we are anxious to see what users are going to develop with these new capabilities.
There are already many Raspberry Pi users who embedded Hailo AI accelerators into their projects before we launched this product. One of them is Velo.ai who is using a Raspberry Pi Compute Module 4 empowered by a Hailo-8 AI accelerator to provide the Copilot, a camera-based, high-performance, low-power rider assistance system to improve bicycle riders' road safety.
Eben Upton (Raspberry Pi): Based on past experience, I would expect to see our hobbyists finding new ways to apply the usual range of "standard" models, including object detection and pose extraction. This represents a continuation of an existing trend, but with the Hailo accelerator supporting higher resolutions and frame rates, and more modern, higher-accuracy models.
Also: The best Raspberry Pi alternatives
In the industrial space, there are numerous possibilities, including autonomous control of industrial robotics, and automated product inspection.
But inevitably the most exciting applications are the ones we can't predict. Just as Raspberry Pi drove down the cost of high-performance embedded compute, the AI Kit drives down the cost of edge AI inference acceleration. This means that many novel applications, which today would be cost-prohibitive, will have a positive return on investment. It's going to be exciting to discover what those applications are.
ZDNET: Can you explain how the AI capabilities of the Raspberry Pi AI Kit can be utilized in home automation projects?
Orr Danon (Hailo): As mentioned earlier, we can only imagine how users would leverage the advanced AI capabilities of their Raspberry Pi. This is basically empowering the house with vision and intelligence; that could be applied into a whole range of applications, from feeding pets to watering plants, creating new games or empowering the home assistant with eyes and understanding ("suggest a recipe based on the ingredients on my table").
ZDNET: How do you envision the Raspberry Pi AI Kit being used in security applications?
Orr Danon (Hailo): The AI Kit allows Raspberry Pi users to embed advanced video analytics to identify people, animals, objects and vehicles. Users can develop access control and perimeter protection applications based on these capabilities.
Eben Upton (Raspberry Pi): I think this is one of the use cases where the AI Kit will really shine. We already see many examples of security or presence-detection applications from both enthusiasts and industrial customers: in fact, the "Santa Detector" was one of the earliest educational projects featured on our website. Adding a layer of semantic understanding of the scene will allow the user to experiment with more sophisticated responses to specific visual cues.
ZDNET: What advantages does the Hailo-8L provide for robotics projects compared to other processors?
Orr Danon (Hailo): In comparison to other processors, the Hailo-8L provides high compute power with a double-digit TOPS [Theoretical Operations Per Second] capacity, at the industry's best performance-to-cost and performance-to-power consumption ratios.
This enables the AI kit to provide very attractive computing capabilities while maintaining cost efficiency.
ZDNET: How does the integration of AI with Raspberry Pi enhance its capabilities for educational purposes?
Eben Upton (Raspberry Pi): We've always believed that it is important to teach children about the state-of-the-art in computer science and technology. We do them a disservice by teaching them a simplified or obsolete view of the world.
Also: How to install Linux on your Raspberry Pi
The AI Kit allows us to open their eyes to the new tools that neural network models add to the toolbox that they'll take with them into their careers.
ZDNET: What kind of AI models can be deployed on the Raspberry Pi AI Kit, and how can users create or obtain them?
Orr Danon (Hailo): [The AI Kit] is fully integrated with Raspberry Pi's camera software stack and supports numerous out-of-the-box AI applications through Hailo's robust software suite and model zoo which covers all the popular machine vision tasks, such as classification, object detection, semantic segmentation, pose estimation, depth estimation super resolution, image denoising, etc.
ZDNET: How does the Raspberry Pi AI Kit address the growing demand for edge AI solutions?
Eben Upton (Raspberry Pi): The AI Kit provides a ready-to-use platform for developing and deploying "edge AI" applications. With our Compute Module range of system-on-module products, we support seamless transition from low-to-medium volume deployments using our Raspberry Pi single-board computer products and AI Kit, to higher volume developments using Compute Module accompanied by Hailo 8 or 8L silicon on a carrier board.
ZDNET: What security measures are in place to ensure the safety and privacy of AI applications developed with the Raspberry Pi AI Kit?
Orr Danon (Hailo): Since the AI is running on the edge device and not in a cloud server, this solution provides the highest level of privacy and data safety. Personally identifiable information (PII) can be processed on the edge, and anonymized information can be uploaded to the cloud for further analysis and storage.
Eben Upton (Raspberry Pi): It is the responsibility of the developer to ensure that their application ensures the safety and respects the privacy of its users. But, as Orr says, locating inference operations at the edge of the network reduces the scope for privacy violations, and, by enabling the edge device to autonomously make AI-supported decisions in the absence of network connectivity, eliminates certain opportunities for safety issues to arise.
ZDNET: How does the Raspberry Pi AI Kit contribute to the democratization of AI technology?
Eben Upton (Raspberry Pi): We believe very strongly in the social value of giving startups and small businesses access to cutting-edge technology. In support of this mission, and as with all Raspberry Pi products, the AI Kit is available to all comers at a low, fixed, price point [US$70].
Orr Danon (Hailo): With an attractive price tag and easy-to-use software package this product allows developers around the world, whether professionals or hobbyists, to apply intelligence into their projects, combining the power of the Raspberry Pi ecosystem with the new world of applications enabled by AI.
ZDNET: What advice would you give to new users starting out with the Raspberry Pi AI Kit?
Eben Upton (Raspberry Pi): Explore the wide range of curated, pre-trained, pre-converted models in the Hailo model zoo for inspiration. Many common use cases can be supported using these models, or alternatively they can serve as inspiration and a jumping off point for your own experimentation.
Orr Danon (Hailo): Let your imagination run wild…almost everything is possible.
ZDNET: Any final thoughts or insights you'd like to share with our readers?
Orr Danon (Hailo): Hailo is very excited about this cooperation. For Hailo, this is an opportunity to reach new audiences and truly enable AI at the edge for both professional engineers and creative enthusiasts. As generative AI is slowly expanding from the cloud to the edge, we see a lot of potential for future collaboration with Raspberry Pi and new ways to empower developers.
Eben Upton (Raspberry Pi): This is a watershed moment for us: the first time we've deployed a first-party AI product, and an opportunity to work with a partner whom we've admired for a long time. We can't wait to see what people do with the AI Kit, and where our collaboration with Hailo takes us.
Final thoughts
ZDNET's editors and I would like to give a huge shoutout to Eben Upton and Orr Danon for taking the time to engage in this in-depth interview. There's a lot of food for thought here. Thank you, Eben and Orr!
Also: Don't buy a Raspberry Pi 5 without also buying this amazing accessory
The Raspberry Pi 5 and the AI Kit are available from Raspberry Pi's list of approved resellers.
What do you think? Did their recommendations give you any ideas about how to engage with AI in your creator journey, or for your company or organization? Let us know in the comments below.
