#############? AI compliance, privacy, security, copyright ###########
- The company used primarily public posts rather than private conversations.
- "We've tried to exclude datasets that have a heavy preponderance of personal information,” Meta President of Global Affairs Nick Clegg told Reuters at this week's Meta Connect conference. Clegg adds that the “vast majority” of the data Meta used to train its AI assistants is publicly available.?
- Earlier this week, Meta announced a line of AI chatbots across its messaging platforms including WhatsApp, Instagram, and Messenger. The chatbot can help with things like answering questions, planning trips, and generating images from text prompts. The assistant will also be able to provide real-time results from Microsoft’s Bing search.
- Meta’s model was based partially on its LLaMA 2 language model, which was released in July.
- Many companies have recently come under fire for using web content to train their AI without the permission of the original author or artist. That includes Elon Musk's xAI, which plans to train its AI on tweets, prompting artists who share their content on Twitter to threaten to leave the site.
- Several authors have been left shocked after seeing artificial intelligence (AI)-generated ripoffs of their books being sold on e-commerce giant Amazon (NASDAQ:AMZN). Writer-journalist Rory Cellan-Jones recently saw a memoir on Amazon he wrote with a "naively designed cover by someone he had never heard of," reports The Guardian.
- In August, author Jane Friedman, who writes about publishing, forced Amazon to remove five bogus titles in her name that were AI-generated.
- "Amazon is clearly facing significant challenges with the influx of AI-generated products in its stores, and it appears to be playing catchup," Nicola Solomon, the chief executive of the Society of Authors (SoA), was quoted as saying in the report.
- In the same month, authors Michael Chabon, David Henry Hwang, Rachel Louise Snyder and Ayelet Waldman alleged in a lawsuit that OpenAI benefits and profits from the "unauthorised and illegal use" of their copyrighted content.
- This significant release comes on the heels of Bedrock’s debut, where it quickly proved its worth by meeting the demands of enterprise users. Moreover,?in July, AWS upgraded the suite by including additional models, like?Anthropic Claude 2?and?Stability AI SDXL 1.0?models and the essential proprietary?Titan Embeddings model.
- Among the many things that AWS has improved in Amazon Bedrock is ensuring compliance with regulatory standards, with a particular focus on the European Union’s General Data Protection Regulation (GDPR).?Vasi Philomin, the Vice President and General Manager for Generative AI at Amazon, has?affirmed?this compliance, highlighting the company’s commitment to safeguarding data privacy and protection.
- Although OpenAI’s most powerful generative AI models are available through?Microsoft Azure, access to these models remains highly restricted. Developers keen on using them must undergo a stringent application process, providing detailed use cases for consideration. Furthermore, even after gaining access, all solutions that deploy Azure OpenAI models must go through a use case review before being approved for production use. This cumbersome process stands in stark contrast to what AWS offers, making it a better choice for developers and the collaboration with Hugging Face further simplifies the process.
- AI Guardrails, released Sept. 27, can be integrated into any generative AI product and positioned between the large language model and the end user.?Aporia said it ensures fair and responsible usage by eliminating discriminatory or inappropriate LLM and chatbot responses depending on the standards of the workplace. It’s also intended to reduce the risk of data leakage and protect against the inadvertent disclosure of sensitive information such as credit card or medical data, a feature that the company said guarantees user safety and helps optimize performance.
- An online?survey?by Tidio LLC found that 86% of the nearly 1,000 people who responded have personally experienced hallucinations and 46% said they encounter them frequently. For example, asking ChatGPT about the record for crossing the English Channel on foot generates a response even though such an achievement is impossible.
- Aporia said its technology encompasses observability, visibility, detection and control to promote the responsible and secure integration of AI technology into various scenarios. Real-time alerts notify organizations about potential issues related to AI performance and unified visibility provides a transparent and consolidated view of all LLM operations. This allows organizations to proactively analyze model behavior to stay ahead of hallucinations.
- Microsoft's?Bing Chat?AI chatbot has gained much popularity because it is free, connected to the internet, powered by?GPT-4, and multimodal, boasting features the accessible version of?ChatGPT?doesn't have. However, a new report shows that its responses may contain malware.?
- Reports by?Malwarebytes?show that those ads are now being used for malvertising, the practice of using online advertisements that would interest users to spread malware and infiltrate unsuspecting users' devices.?
- According to the report, scammers are inserting malicious ads so that they appear first over the organic result ad in the hover ad experience in Bing Chat.?
- The solution to avoiding this attack isn't necessarily to stop using Bing Chat, as malicious ads can be anywhere online. Instead, users should be mindful before downloading things off the web to their devices and ensure the site is legitimate first.?
- The Federal Court of Appeal in the USA has just ruled that Google is not covered by exemption for journalistic or artistic work. In a 2-1 court ruling, Google which drives more than 75% of internet searches in Canada, which opens the door for people to demand that their names in any articles are made unsearchable known as the right to be forgotten.
- More importantly it signals the imperative for International Human Rights Laws to align to protect our privacy as we enter a world which is increasingly like the Minority Report, where everything one does is known and never forgotten.
- Google Bard, the search giant’s conversational AI product, underwent a?big update?last week that earned?mixed reviews. But this week, another, older Bard feature is coming under scrutiny:?SEO consultant Gagan Ghotra?observed that Google Search had begun to index shared Bard conversational links into its search results pages, potentially exposing information users meant to be kept contained or confidential.
- This means that if a person used Bard to ask it a question then shared the link with a designated third-party, say, their spouse, friend or business partner, the conversation accessible at that link could in turn be scraped by Google’s crawler and show up publicly, to the entire world, in its Search Results.
- “Most users wouldn’t be aware of the fact that shared conversation mean it would be indexed by Google and then show up in SERP, most people even I was thinking of it as a feature to share conversation with some friend or colleague & it being just visible to people who have conversation URL.”
- Researchers have demonstrated that knowledge workers significantly degrade the quality of their work by?asking?OpenAI's GPT-4?technology?to perform tasks it didn't train for, and then failing to spot "hallucinations" in its answers.
- A new?study of more than 750 strategy consultants?showed that AI helped them?produce better content,?more quickly in many tasks — but the consultants were "less likely to produce correct solutions" by attempting tasks of similar difficulty which fell outside the AI model's capabilities.
- Highly skilled workers struggled to navigate what AI can and can't do well, and a reliance on clichéd GPT-4 outputs reduced the group's diversity of thought by 41%.
- AI can quickly and cheaply increase worker performance, but taking shortcuts with AI training might cause more mistakes, which could potentially blow up in an employer's face.
- Even the AI experts behind this report are stumped by "novel and unexpected" AI capabilities and capability gaps in the workplace.
################# AI future ######################
- Manufacturing is equally benefiting from predictive maintenance powered by deep learning. Machines equipped with AI can predict when they need maintenance, reducing downtime by up to 50% and increasing efficiency. It's a game-changer for industries where every minute of operation counts.
- Investors are also turning to deep learning to decode market complexities. AI systems analyze vast datasets, news sentiment, and market trends to make informed investment decisions. These algorithms can process information at speeds impossible for humans, giving them an edge in today's fast-paced financial markets.
- Deep learning models are being used to detect diseases from medical scans with unprecedented accuracy. They can spot abnormalities in X-rays, MRIs, and CT scans that might elude the human eye. In some studies, AI systems have shown diagnostic accuracy rates exceeding 90%.
- Customer service chatbots can understand and respond to the emotional nuances of human communication, providing a more human-like interaction. This not only enhances customer satisfaction but also frees up human agents for more complex tasks. As a result, some companies report a 70% reduction in response time and a 50% increase in customer satisfaction.
- No matter where you are, AI, ChatGPT and related tools are the most popular topics of conversation. Statistics show that the global generative AI market is growing at a?CAGR of just over 27%?and will surpass $22 billion by 2025.
- We have learned that AI is great for aggregating, parsing and analyzing data. It’s also a good frontline customer service stand-in, providing round-the-clock basic assistance for customers in nearly every industry.?Gen AI?for a more “human” first-line customer service experience has already been tested with positive results, and it’s a simple, effective way for other companies to “get their feet wet” with the technology.
- AI will be the “multivitamin supplement” of many industries. It won’t replace humans in the same way that supplements don’t replace a healthy diet. Still, it will strengthen companies’ existing operations and fill in the gaps that are currently making work more burdensome for human laborers.
- Similar benefits of human-AI collaboration can be seen in healthcare, with AI that can be trained to assist patients with?recovery treatments?or?perform routine tasks?in medical offices or hospitals, freeing nurses and doctors up to focus on patient outcomes. It’s present in warehouses, manufacturing and q-commerce as well, and I think the future will see even more cooperative roles that we don’t even have a frame of reference for yet.?
- The most likely future for AI is treating it as a “joint effort” technology that requires humans to reach its full potential, so these types of collaborative applications are what corporations should focus on when considering new AI integrations.
- New reseach predicts that Switzerland and Sweden will be among the biggest beneficiaries
- Using 40-sub-indicators, the researchers analysed their capacities for AI innovation, diffusion of the AI effectively, and adaptability to the impacts. The three categories were then combined to calculate a global ranking.
- Britain may be further impacted by its position outside of the EU — both positively and negatively. While Capital Economics doesn’t expect Brexit to be a major factor in the AI economy, the research team can envision some effects.
- “On the plus side, the UK could benefit from lighter-touch regulation if the EU regulates AI more strictly in future,” Andrew Kenningham, the firm’s Chief Europe Economist, told TNW. “But set against that, there could be reduced collaboration on AI projects or the UK could be hindered from participation in large-scale European AI initiatives.”
- “Europe is a long way behind the US and China on our sub-index of ‘innovation’ essentially because it invests less in AI research and has less academic research in that area,” Kenningham explained.
#################### AI products, offering ####################
- Preston Gralla?of Computerworld gave a run-down of the newly-released Copilot – Microsoft’s “everyday AI companion.” After a trial run, Gralla concluded that?Copilot?is efficient at completing basic tasks and can search for/retrieve valuable information in minutes. However, he emphasized that “Copilot also doesn’t give Windows any new capabilities.”
- It also produced inaccurate guidance with other commands. Gralla’s takeaway is that Copilot is indeed helpful with simplifying existing features – but doesn’t deliver on more complicated tasks.
- Gralla is right. It’s still very early days for all AI, including Microsoft’s offerings. I’m excited to see how Copilot in Office 365 will improve productivity by creating docs and presentations and even allowing me to attend more than one meeting at a time but let’s not be na?ve…there’s still far to go before this stuff is really working accurately and reliably. The big question is whether or not users will pay $30 per month for such a young product.
- Whisper is an automatic speech recognition system that demonstrates incredible accuracy in understanding spoken words. It was built by OpenAI, presumably for use in systems like?ChatGPT, where you can now?converse with an AI, but the company also open-sourced Whisper so that the community could use it as well.
- As for how it compares to other tools, OpenAI says that Whisper makes up to 50% fewer errors than other language models, and I believe it. I have used a lot of tools over the years to try and transcribe audio, and nothing has been as accurate as Whisper for me. As I mentioned, I transcribed a 25-minute interview that came out flawlessly, which pretty much every tool struggles with.
- The one thing particularly interesting about Whisper is that it's not a tool aimed at end users but rather at developers and researchers. OpenAI said the reason for open-sourcing the models and code was to "serve as a foundation for building useful applications and for further research on robust speech processing." You can still set it up and use it, but it's not really a consumer product yet.
- Being at the forefront of cost reduction and efficiency enhancement for large models, the?Colossal-AI?team maximizes the core capabilities of LLaMA-2. Through innovative training techniques, Colossal-AI has achieved remarkable results by utilizing only approximately?0.0085 trillion tokens of data, investing 15 hours, and incurring training costs in the range of a few hundred dollars. This strategy has yielded a high-performance Chinese LLaMA-2 model that consistently outperforms competitors across multiple evaluation benchmarks.
- In contrast to the original LLaMA-2, Colossal-AI's model not only enhances Chinese language capabilities but also further refines its proficiency in English. Remarkably, it exhibits performance levels that rival state-of-the-art (SOTA) models of similar scale within the open-source community.
- Underpinning Colossal-AI's approach are steadfast open-source principles. As a result, this model is made?accessible without any commercial restrictions, with complete transparency extended to the entire training process, code, and model weights. In conjunction with this, Colossal-AI offers the comprehensive evaluation framework, ColossalEval, facilitating cost-effective reproducibility.
- Moreover, the methodologies developed by Colossal-AI can be?readily applied across various domains, facilitating the economical construction of large models that are pre-trained from scratch.
- AI Copilots can learn from the developer’s coding patterns and adapt to their preferences over time, which enhances the user’s assistance experience. Well-known AI Copilots include GitHub Copilot and OpenAI GPT-3. AI Copilots leverage various artificial intelligence, natural language processing (NLP), machine learning, and code analysis. AI Copilots are often updated regularly to incorporate new programming languages, frameworks, and best practices, ensuring they remain valuable to developers as technology evolves.
- Now, a team of researchers design?OpenCopilot. It is a user’s own AI copilot, trained specifically for their product and their requirement. Unlike generic AI models, OpenCopilot deeply integrates with a product’s underlying APIs by the primary function and effortlessly executes API calls whenever required. It uses LLMs to determine if the user’s request requires calling an API endpoint. It stands as a tool that can significantly improve efficiency and reduce the manual work involved in interfacing with APIs.?
- OpenCopilot can call your underlying APIs and transform the responses into meaningful texts. It can also automatically produce certain request payload fields based on the context. Users need to provide their API/backend definition as well as their public endpoints to call them. Users can also embed OpenCopilot’s chat bubble into their SaaS applications. OpenCopilot ensures the provided schema is valid to produce optimized results.
- Genpact?has announced an expanded collaboration with Amazon Web Services (AWS) aimed at revolutionising financial crime risk operations through the integration of generative AI and LLMs. The partnership entails the integration of Genpact’s cloud-based financial crime suite,?riskCanvas, with?Amazon Bedrock, yielding significant efficiencies and benefits for clients, including Apex Fintech Solutions.
- Leveraging its existing relationship with AWS, Genpact is merging its intellectual assets and industry expertise with AWS’s generative AI capabilities. The seamless integration of Amazon Bedrock FMs into Genpact’s riskCanvas financial crimes software suite aims to unlock substantial value, enhance speed, and improve accuracy in detecting, investigating, and preventing financial crime threats for businesses.
- Bedrock's unique differentiator is its granular focus on enterprise requirements for compliance and observability. The service checks all the boxes in terms of GDPR compatibility and further integrates seamlessly with Amazon CloudWatch, enabling full-scope logging functionalities. Despite its myriad advantages, it is worth noting that Bedrock faces stiff competition, notably from Google's Vertex AI. What gives AWS an edge is Bedrock’s deep integration within the expansive AWS ecosystem, often considered both a strategic and perceived advantage, particularly dependent on a customer’s existing cloud framework.
- The tech giant says its new platform, Cloudflare AI, allows developers to build scalable AI applications quickly and affordably.
- Cloudflare’s new AI platform—released on the company's 13th anniversary—is intended to enable developers to build “full-stack” and “production-ready” AI applications on Cloudflare's network rather than their own servers, providing a fast and affordable way to deploy AI models without the hassles of direct infrastructure management.
- Other companies providing cloud AI deployment services include Amazon Web Services (AWS), Azure by Microsoft, IBM Cloud, and Google Cloud.
- Companies working with Cloudflare to develop its AI platform include Microsoft, NVIDIA, Hugging Face, Databricks, and Facebook parent company Meta.
- “AI is shaping the way we build the metaverse,” said Chief Technology Officer Andrew Bosworth.
- Meta executives laid down a robust blueprint of how?AI will be the linchpin?in many company initiatives—including crafting the metaverse.
- Meta also did not skimp on safety. The company introduced a LLaMA-2 “red team” that aims to button down the company’s AI language mode, which is especially critical since its open-source nature could create a Pandora's box of uncensored content.
- To scale its AI ambitions, Meta announced a new partnership with Amazon Web Services (AWS) to let AWS customers run Llama2 with just an API key, a move that significantly lowers the barrier for developers keen on exploring the AI domain. Meanwhile, Meta’s new AI Studio is a toolkit for developers to create AI products across various domains including fashion, design, and gaming.
- Europe’s startup contribution to the generative AI bonanza, Mistral, has released its first model. Mistral 7B is free to download and be used anywhere — including locally.?
- French?AI?developer Mistral says its Large Language Model is optimal for low latency, text summarisation, classification, text completion, and code completion. The?startup?has opted to release Mistral 7B under the Apache 2.0 licence, which has no restrictions on use or reproduction beyond attribution.
- The 7.3 billion parameter Mistral 7B reportedly outperforms larger models, such as Meta’s 13 billion parameter Lama 2, and requires less computing power. Indeed, according to its developers, it “outperforms all currently available open models up to 13B parameters on all standard English and code benchmarks.”?