https://adgully.me/post/5101/surge-seen-in-ai-powered-startups-new-cybersecurity-tools

Surge seen in AI-powered startups, new cybersecurity tools

In 2024, there will be a surge in AI-driven startups, marked by the development of cybersecurity tools leveraging generative AI models and the integration of artificial intelligence as a supportive assistant.In its prediction for 2024, Positive technologies, a cybersecurity leader, sees setting up of services with the help of natural languages, private blockchain, trusted blockchain, and advanced encryption methods, which are fuelling the evolution of blockchain technology.2024 is set to feature an astonishing landscape of change for organisations across the Middle East especially in light of the region’s race towards digitalisation.Artificial intelligence and cybersecurityAlexandra Murzina, Head of Advanced Technologies, Positive Technologies, says new capabilities of generative AI have reshaped many industries this year. Just a year ago, few would have imagined the multi-pronged uses of AI: image generation, text generation along with text-to-speech, as well as scriptwriting and the creation of script-based videos in real time.“Today, we are witnessing a boom of AI-powered startups. Artificial intelligence has advanced so fast that it has taken root in some professional areas, forcing the workforce to master new tools.”This trend also applies to cybersecurity. AI technologies have benefited both defence teams and attackers. They have helped companies make threat detection and response processes more effective, and cybercriminals improve their cyber kill chains. GitHub now features many different cybersecurity tools based on generative AI models.AI as assistant“In addition, we are seeing a trend towards the use of artificial intelligence as an assistant. Take Microsoft Security Copilot, for example. This trend is pushing the industry to rethink conventional cybersecurity products.”The open-source community has contributed a lot to this progress by developing many of the ML solutions currently available. Companies, academies, and individual enthusiasts have trained many large language models, shared their image generation algorithms based on the Stable Diffusion model, and posted useful materials and tutorials. This has made advanced technologies more accessible than ever before.Knowledge distillation of large language models was a major trend of 2023. Training LLMs takes a lot of effort and using them is expensive. That's why experts are looking for a solution that would simplify this task and make it available to more people. This has resulted in quantified and distilled LLMs that can run on laptops, and some even on mobile devices. At the same time, guidelines and recommendations on performing these tasks are now available.Creating servicesAnother emerging trend is creating services with the help of natural languages. It is projected to make big strides as early as next year, enabling development of breakthrough technologies without coding. For example, prior to AI, if you wanted to build an application or query a database, you needed to know the language syntax to work with those tools. Now you can do it by writing prompts in natural languages, and LLMs will follow your instructions and apply the necessary format.Moreover, the industry's still trying to figure out what an artificial general intelligence (AGI) should be. AGI is a hypothetical type of artificial intelligence capable of performing any intellectual task even better than human professionals. There are many opinions and interpretations. In November 2023, for example, Google DeepMind published a paper presenting their vision of artificial general intelligence as a framework for classifying the capabilities and behavior of AGI models and their precursors.Growth drivers for blockchain projects in 2024In 2023, the world witnessed a rapid advancement of technology based on zero-knowledge proof (ZKP). ZPK technology already found application in a range of Ethereum L2 chains and Proof-of-Reserves approaches.It even started spreading beyond blockchain. Much work has been done in the field of identification. For example, the Worldcoin project, which was launched not so long ago, uses retinal scans to identify individuals and has already gained over one million users. Inter-blockchain technologies, such as the Cosmos and Polkadot ecosystems, will definitely continue to expand.The blockchain market is growing, indicating a rise in investment and innovation spending in this area. In addition to cryptocurrencies, blockchain is being widely used in industry and the Internet of Things. Since 2020, we've observed strong investment in blockchain projects, which is driving the application of these technologies in business.The geographic distribution of blockchain increases the sustainability of projects: North America and Asia-Pacific are leaders in the use of blockchain technology. The emerging standards and solutions such as private blockchain, trusted blockchain, and advanced encryption methods, are fuelling the evolution of blockchain technology.
https://adgully.me/post/4350/amazon-trains-massive-ai-model-olympus-to-rival-openai-alphabet

Amazon trains massive AI model 'Olympus' to rival OpenAI, Alphabet

Amazon has formed a dedicated team to train a massive AI model called "Olympus," according to a Reuters report.Amazon is making a significant financial investment in this large language model (LLM) with the hope that it can compete with leading models developed by OpenAI and Alphabet, according to insider information.The "Olympus" model is said to possess a staggering 2 trillion parameters, potentially making it one of the largest models under development. In comparison, OpenAI's renowned GPT-4 model is reported to have one trillion parameters.Rohit Prasad, the former head of Alexa, is leading the team responsible for "Olympus" and now reports directly to CEO Andy Jassy. In his role as the head scientist for general artificial intelligence (AGI) at Amazon, Prasad has brought in researchers who were previously engaged with Alexa AI and the Amazon science team to collaborate on model training, thus consolidating AI efforts across the company with dedicated resources.Already Amazon has trained smaller models, such as Titan, and has also established partnerships with AI model startups like Anthropic and AI21 Labs, making these models available to users of Amazon Web Services (AWS).The insiders indicate that there is no specific timeline for the release of the new model.LLMs serve as the foundational technology for AI tools that learn from extensive datasets to generate responses akin to those of humans. Training larger AI models is a costly endeavor due to the substantial computing power required. In an earnings call in April, Amazon executives expressed their intention to increase investments in LLMs and generative AI while reducing expenditures in their retail business related to fulfillment and transportation.