Categories
News

La aplicación web Gemini obtiene soporte para los complementos Google Keep y Google Tasks

[ad_1]

mellizo La web está recibiendo soporte para nuevos complementos que permitirán a los usuarios sacar más provecho del chatbot de IA. El cliente web del chatbot ahora es compatible con Google Keep y Google Tasks. Estos complementos se lanzaron por primera vez. Liberado Google ha anunciado el lanzamiento de la aplicación Gemini exclusivamente para la serie Pixel 9. Sin embargo, parece que el gigante tecnológico ahora está trabajando para ampliar la función para incluir a más usuarios. Vale la pena señalar que las adiciones en la versión web del chatbot AI están disponibles para todos los usuarios.

Gemini obtiene soporte para nuevos complementos

El personal de Gadgets 360 pudo verificar la presencia de estas adiciones en la versión web de Gemini. La compatibilidad con Google Keep y Google Tasks se puede ver dentro del complemento Workspace, que también permite el acceso a otras aplicaciones como Gmail, Docs, Drive y más.

Para activar la extensión, los usuarios deberán abrir Gemini en el navegador. Una vez en la página web, los usuarios deberán hacer clic en el ícono de Configuración ubicado en la esquina inferior izquierda. Allí, los usuarios pueden hacer clic en Complementos y encontrar… Espacio de trabajo de Google opción y actívela. Sin embargo, la extensión solo funcionará si el usuario ha iniciado sesión en su cuenta de Google Workspace en el navegador.

Después de activar la extensión, los usuarios pueden regresar a la interfaz principal y escribir “@” en el campo de texto seguido del nombre de la extensión para usar la función. Por ejemplo, Google Keep permite a los usuarios crear notas y listas para diferentes propósitos. El proceso es sencillo. Los usuarios pueden pedirle a la IA que le dé un nombre a la lista y los elementos que se colocarán en ella.

Los usuarios también pueden agregar recomendaciones recibidas por AI directamente a la lista. Pero Gemini no puede editar, eliminar, compartir una nota existente, agregar fotos ni verlas en una nota.

Asimismo, Google La extensión Tareas permite a los usuarios usar Gemini para agregar recordatorios y tareas. Al indicar la fecha, la hora y la tarea, la IA puede crear automáticamente un recordatorio para ello. Estas tareas también se pueden configurar durante una conversación con el chatbot pidiéndole que “Agregue un recordatorio”. Géminis podrá comprender y agregar contexto.

[ad_2]

Source Article Link

Categories
Life Style

AI now beats humans at basic tasks — new benchmarks are needed, says major report

[ad_1]

Artificial intelligence (AI) systems, such as the chatbot ChatGPT, have become so advanced that they now very nearly match or exceed human performance in tasks including reading comprehension, image classification and competition-level mathematics, according to a new report (see ‘Speedy advances’). Rapid progress in the development of these systems also means that many common benchmarks and tests for assessing them are quickly becoming obsolete.

These are just a few of the top-line findings from the Artificial Intelligence Index Report 2024, which was published on 15 April by the Institute for Human-Centered Artificial Intelligence at Stanford University in California. The report charts the meteoric progress in machine-learning systems over the past decade.

In particular, the report says, new ways of assessing AI — for example, evaluating their performance on complex tasks, such as abstraction and reasoning — are more and more necessary. “A decade ago, benchmarks would serve the community for 5–10 years” whereas now they often become irrelevant in just a few years, says Nestor Maslej, a social scientist at Stanford and editor-in-chief of the AI Index. “The pace of gain has been startlingly rapid.”

Speedy advances: Line chart showing the performance of AI systems on certain benchmark tests compared to humans since 2012.

Source: Artificial Intelligence Index Report 2024.

Stanford’s annual AI Index, first published in 2017, is compiled by a group of academic and industry specialists to assess the field’s technical capabilities, costs, ethics and more — with an eye towards informing researchers, policymakers and the public. This year’s report, which is more than 400 pages long and was copy-edited and tightened with the aid of AI tools, notes that AI-related regulation in the United States is sharply rising. But the lack of standardized assessments for responsible use of AI makes it difficult to compare systems in terms of the risks that they pose.

The rising use of AI in science is also highlighted in this year’s edition: for the first time, it dedicates an entire chapter to science applications, highlighting projects including Graph Networks for Materials Exploration (GNoME), a project from Google DeepMind that aims to help chemists discover materials, and GraphCast, another DeepMind tool, which does rapid weather forecasting.

Growing up

The current AI boom — built on neural networks and machine-learning algorithms — dates back to the early 2010s. The field has since rapidly expanded. For example, the number of AI coding projects on GitHub, a common platform for sharing code, increased from about 800 in 2011 to 1.8 million last year. And journal publications about AI roughly tripled over this period, the report says.

Much of the cutting-edge work on AI is being done in industry: that sector produced 51 notable machine-learning systems last year, whereas academic researchers contributed 15. “Academic work is shifting to analysing the models coming out of companies — doing a deeper dive into their weaknesses,” says Raymond Mooney, director of the AI Lab at the University of Texas at Austin, who wasn’t involved in the report.

That includes developing tougher tests to assess the visual, mathematical and even moral-reasoning capabilities of large language models (LLMs), which power chatbots. One of the latest tests is the Graduate-Level Google-Proof Q&A Benchmark (GPQA)1, developed last year by a team including machine-learning researcher David Rein at New York University.

The GPQA, consisting of more than 400 multiple-choice questions, is tough: PhD-level scholars could correctly answer questions in their field 65% of the time. The same scholars, when attempting to answer questions outside their field, scored only 34%, despite having access to the Internet during the test (randomly selecting answers would yield a score of 25%). As of last year, AI systems scored about 30–40%. This year, Rein says, Claude 3 — the latest chatbot released by AI company Anthropic, based in San Francisco, California — scored about 60%. “The rate of progress is pretty shocking to a lot of people, me included,” Rein adds. “It’s quite difficult to make a benchmark that survives for more than a few years.”

Cost of business

As performance is skyrocketing, so are costs. GPT-4 — the LLM that powers ChatGPT and that was released in March 2023 by San Francisco-based firm OpenAI — reportedly cost US$78 million to train. Google’s chatbot Gemini Ultra, launched in December, cost $191 million. Many people are concerned about the energy use of these systems, as well as the amount of water needed to cool the data centres that help to run them2. “These systems are impressive, but they’re also very inefficient,” Maslej says.

Costs and energy use for AI models are high in large part because one of the main ways to make current systems better is to make them bigger. This means training them on ever-larger stocks of text and images. The AI Index notes that some researchers now worry about running out of training data. Last year, according to the report, the non-profit research institute Epoch projected that we might exhaust supplies of high-quality language data as soon as this year. (However, the institute’s most recent analysis suggests that 2028 is a better estimate.)

Ethical concerns about how AI is built and used are also mounting. “People are way more nervous about AI than ever before, both in the United States and across the globe,” says Maslej, who sees signs of a growing international divide. “There are now some countries very excited about AI, and others that are very pessimistic.”

In the United States, the report notes a steep rise in regulatory interest. In 2016, there was just one US regulation that mentioned AI; last year, there were 25. “After 2022, there’s a massive spike in the number of AI-related bills that have been proposed” by policymakers, Maslej says.

Regulatory action is increasingly focused on promoting responsible AI use. Although benchmarks are emerging that can score metrics such as an AI tool’s truthfulness, bias and even likability, not everyone is using the same models, Maslej says, which makes cross-comparisons hard. “This is a really important topic,” he says. “We need to bring the community together on this.”

[ad_2]

Source Article Link

Categories
Featured

Samsung is going after Nvidia’s billions with new AI chip — Mach-1 accelerator will combine CPU, GPU and memory to tackle inference tasks but not training

[ad_1]

Samsung is reportedly planning to launch its own AI accelerator chip, the ‘Mach-1’, in a bid to challenge Nvidia‘s dominance in the AI semiconductor market. 

The new chip, which will likely target edge applications with low power consumption requirements, will go into production by the end of this year and make its debut in early 2025, according to the Seoul Economic Daily.

[ad_2]

Source Article Link

Categories
News

Claude 3 beats ChatGPT in a variety of different tasks

Claude 3 beats ChatGPT in a variety of different tasks

Just as we thought the artificial intelligence (AI)  model arena was settling down a little, Anthropic has launched Claude 3 which is capable of outperforming ChatGPT in a number of areas. Claude 3 is available in three distinct models: Haiku, Sonnet, and Opus, each offering its own unique capabilities. Opus is the most powerful and capable, designed for complex logic and intense prompts, while Haiku is the fastest but less accurate, intended for instant customer service responses. Sonnet is a free, intermediate model available to the public. Although all are now being tested against ChatGPT and Google Gemini 1.0 Ultra, with Claude 3 Opus outperforming both in various benchmarks.

Let’s dive into the details. Anthropic has three models as each has its own strengths. Haiku is the speedster, designed for quick replies, perfect for customer service. Sonnet is the middle child, offering a good mix of speed and smarts, and it’s free for you to use, although there’s a limit on how much you can use it each day. Then there’s Opus. Opus is the star of the show, the one you’d want on your team for the toughest challenges. It’s not free, costing $20 a month, but for that price, you get the best logic and problem-solving abilities out there.

Now, let’s talk about a feature that’s really exciting: vision capabilities. Opus can now understand pictures. This is a big deal because it opens up so many possibilities. Think about it: you could show it a design, and it could give you feedback, or you could ask it to describe what’s happening in a photo. This is a huge step forward for AI.

Claude 3 results beat ChatGPT

But it’s not just about being smart and fast. Anthropic has worked hard to make sure these AIs are user-friendly and less biased. This is important because it means more people can use them without running into unfair or prejudiced responses. Making AI accessible and fair is a big challenge, but it’s one that Anthropic is tackling head-on.

To really understand how good these AIs are, Anthropic didn’t just use standard tests. They made their own benchmarks. This means they could really focus on what each AI model is best at and make sure they’re up to the task, no matter what industry they’re used in. Watch the overview video below created by Matt Wolfe to learn more about the differences between the three different Claude AI models now available and how their results compare to OpenAI’s ChatGPT.

Here are some other articles you may find of interest on the subject of Claude  3

The differences between Claude 3 AI models

The Claude 3 model family introduces a significant advancement in the domain of artificial intelligence, comprising three distinct models named Haiku, Sonnet, and Opus, each designed to cater to varying demands of speed, intelligence, and application scope. This family of models represents a new benchmark in AI capabilities, offering enhanced performance across a broad spectrum of cognitive tasks. Here’s an overview of each:

Haiku

Haiku is the entry-level model within the Claude 3 family, characterized by its unparalleled speed and cost-effectiveness. It’s designed for applications requiring near-instantaneous responses, making it ideal for tasks like customer interactions, content moderation, and efficient management of logistics and inventory. Despite its rapid response rate, Haiku doesn’t compromise on intelligence, providing smart and accurate support in live interactions, translations, and knowledge extraction from unstructured data. With a context window of 200K tokens and a pricing model designed to be highly affordable, Haiku aims to be the fastest and most cost-efficient option in its intelligence category, catering to use cases that demand swift, reliable AI processing.

Sonnet

Sonnet occupies the middle ground in the Claude 3 model family, offering a balanced mix of intelligence and speed. It is twice as fast as its predecessors (Claude 2 and Claude 2.1) and is designed to support a wide range of enterprise workloads at a lower cost. Sonnet excels in rapid knowledge retrieval, sales automation, data processing, and code generation. Its robust performance makes it suitable for tasks like product recommendations, forecasting, targeted marketing, and quality control. Like Haiku, Sonnet offers a 200K context window but brings enhanced endurance and affordability for large-scale AI deployments, making it an ideal choice for businesses seeking a potent combination of speed, intelligence, and cost-efficiency.

Opus

Opus represents the pinnacle of the Claude 3 model family, offering the highest level of intelligence among the trio. It outperforms other models in the market on complex cognitive tasks, showcasing near-human levels of comprehension and fluency. Opus is capable of navigating open-ended prompts and unseen scenarios with remarkable adeptness, making it suitable for advanced applications like task automation across APIs and databases, R&D, and strategic analysis. Although it delivers similar speeds to Claude 2 and 2.1, its superior intelligence enables it to process complex information and perform highly sophisticated tasks. With a 200K context window, expandable to 1 million tokens for specific cases, Opus is tailored for users requiring the utmost in AI capability.

Cross-Model Features and Innovations

All three models in the Claude 3 family demonstrate significant improvements in vision capabilities, processing a wide range of visual formats and reducing unnecessary refusals by showing a more nuanced understanding of requests. They exhibit enhanced accuracy in responses, with Opus, in particular, showing a remarkable improvement in providing correct answers for challenging questions. The Claude 3 models have been designed with long contexts and near-perfect recall abilities, supporting inputs exceeding 1 million tokens for specific applications.

A key focus has been placed on responsible design, with dedicated teams working to mitigate risks associated with misinformation, CSAM, and other potential harms. Efforts to address biases and promote neutrality have also been advanced, with the Claude 3 models demonstrating reduced biases compared to previous iterations.

In terms of usability, these models have been refined to better follow complex, multi-step instructions and to adhere to brand voice and response guidelines, simplifying the creation of trustworthy customer-facing experiences. The models also support structured output formats like JSON, enhancing their applicability across various natural language processing tasks.

So, what does all this mean for you? If you’re someone who’s interested in AI, whether you’re a developer, a business owner, or just curious, Claude 3’s models are worth paying attention to. They’re not just another set of AIs; they’re a step up in what artificial intelligence can do. With their ability to handle complex tasks, understand images, and do it all with less bias, they’re setting a new standard. And with options for different needs and budgets, there’s a good chance one of these models could be just what you’re looking for.

The Claude 3 model family represents a significant leap forward in the field of AI, offering users a spectrum of choices to meet their specific needs, from the ultra-fast and cost-effective Haiku to the balanced Sonnet, and the highly intelligent Opus. With advancements in speed, accuracy, vision capabilities, and a commitment to responsible AI development, the Claude 3 models set new industry standards and offer promising solutions for a wide range of applications.

Filed Under: Guides, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

How to use Excel Copilot AI assistant to simplify complex tasks

How to use Excel Copilot AI assistant to simplify complex tasks

Microsoft has recently introduced a new feature in Excel that is set to change the way we interact with spreadsheets. This feature, known as Copilot, is an artificial intelligence (AI) tool and assistant that integrates seamlessly into Excel, offering users a more intuitive and efficient way to manage their data. As technology continues to advance, tools like Copilot are indicative of the direction in which software solutions are heading, aiming to make complex tasks more accessible to a broader audience.

To take advantage of Copilot’s full potential, users must have a Microsoft 365 subscription. The Copilot Pro subscription, in particular, unlocks a suite of advanced features designed to enhance productivity. It’s important to note that Copilot is not a standalone application; it works best when files are stored in OneDrive or SharePoint. Additionally, users must format their data into tables before Copilot’s algorithms can be applied effectively.

Microsoft 365

Copilot is integrated into Microsoft 365 in two ways. It works alongside you, embedded in the Microsoft 365 apps you use every day—Word, Excel, PowerPoint, Outlook, Teams, and more—to unleash creativity, unlock productivity, and uplevel skills. Business Chat works across the LLM, the Microsoft 365 apps, and your data—your calendar, emails, chats, documents, meetings, and contacts—to do things you’ve never been able to do before. You can give it natural language prompts like “tell my team how we updated the product strategy” and it will generate a status update based on the morning’s meetings, emails, and chat threads.

Copilot in Excel

Imagine being able to interact with your data by simply asking for what you need in plain language. Copilot can handle a variety of requests, from calculating unique customer counts to adding profit columns, and from creating pivot tables to generating charts. This AI-driven approach is especially beneficial for those who may not be experts in Excel, as it simplifies the use of the program’s sophisticated features. Here are refuted example prompts you can try in Copilot Excel :

  • Give a breakdown of the sales by type and channel. Insert a table.
  • Project the impact of [a variable change] and generate a chart to help visualize.
  • Model how a change to the growth rate for [variable] would impact my gross margin.

Copilot in Excel works alongside you to help analyze and explore your data. Ask Copilot questions about your data set in natural language, not just formulas. It will reveal correlations, propose what-if scenarios, and suggest new formulas based on your questions – generating models based on the questions that help you explore your data without modifying it. Identify trends, create powerful visualizations, or ask for recommendations to drive different outcomes.

Here are some other articles you may find of interest on the subject of Microsoft Copilot AI assistant :

Getting Started with Excel Copilot

1. Requirements:

  • A Microsoft 365 subscription, either a family or a personal plan.
  • An additional subscription to Copilot Pro, granting access to Copilot features across various Microsoft applications including Excel, Word, PowerPoint, and Outlook, along with benefits like using GPT-4 during peak hours and faster image creation with DALL·E 3.

2. Activation:

  • Navigate to the designated website, accessible via a link provided in the video description or a card in the video’s top right-hand corner.
  • Ensure your files are stored in OneDrive or SharePoint as Copilot functions exclusively with cloud-stored files.

Core Features of Excel Copilot

1. Data Analysis and Insights:

  • Formulas and Calculations: Automatically figure out and apply complex formulas based on natural language prompts, significantly reducing the manual effort required in formula creation.
  • Data Visualization: Generate charts and graphs to visually represent data, facilitating easier interpretation and presentation.
  • Highlighting and Sorting: Highlight cells based on specific criteria and sort/filter data seamlessly, enhancing data readability and organization.

2. Efficiency and Productivity Enhancements:

  • Column Addition: Effortlessly add new columns for calculated data, such as profit margins, by simply prompting Copilot with your requirements.
  • Data Conversion: Convert data ranges into tables with a single click, leveraging the advantages of Excel tables like banded rows, quick formatting, and easy data manipulation.
  • Learning and Suggestions: Receive prompt suggestions and sample queries to better engage with Copilot, making the tool accessible to new users and providing inspiration for complex data manipulation tasks.

Practical Applications

1. Simplifying Complex Tasks:

  • Excel Copilot can intuitively understand and execute complex data queries, such as identifying the number of unique customers or calculating total sales per customer, using natural language prompts. This significantly lowers the barrier for performing sophisticated data analysis.

2. Enhancing Data Presentation:

  • The ability to quickly generate and customize charts based on specific data points or trends allows users to present their data in a more impactful manner. Although some customizations may require manual adjustments, Copilot significantly accelerates the initial creation process.

3. Streamlining Data Management:

  • By automating the process of highlighting significant data points, such as high-value transactions, and performing conditional formatting, Copilot aids in quickly identifying key insights within large datasets.

4. Facilitating Advanced Analysis:

  • Copilot can handle requests to analyze data for seasonal trends or outliers, enabling users to identify patterns that may not be immediately apparent through traditional analysis methods.

Limitations and Considerations

While Excel Copilot heralds a new era of data interaction within Excel, it’s important to recognize its current limitations, such as the inability to perform certain customizations directly through AI prompts. Currently, it cannot change chart colors directly, and some power users might notice that its response times are slower than performing tasks manually. Additionally, the tool’s effectiveness is pendant upon clear and precise user prompts, and there may be a learning curve in formulating queries that yield desired outcomes. Despite these initial challenges, the future of Copilot looks promising. It is expected to continue improving and become an indispensable tool for Excel users of all skill levels.

Microsoft Copilot represents more than just a new feature in Excel; it is a step towards making data analysis more democratic and less daunting for users who may not have extensive experience with spreadsheets. As we continue to embrace technological advancements, Copilot is poised to play a significant role in reshaping our interactions with Excel and data management as a whole.

Filed Under: Guides, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

Training AI to use System 2 thinking to tackle more complex tasks

Training AI LLM to use system 2 thinking to tackle more complex tasks

Artificial intelligence seems to be on the brink of another significant transformation nearly every week at the moment, and this week is no exception. As developers, businesses and researchers  dive deeper into the capabilities of large language models (LLMs) like GPT-4, we’re beginning to see a shift in how these systems tackle complex problems. The human brain operates using two distinct modes of thought, as outlined by Daniel Kahneman in his seminal work, “Thinking, Fast and Slow.” The first, System 1, is quick and intuitive, while System 2 is slower, more deliberate, and logical. Until now, AI has largely mirrored our instinctive System 1 thinking, but that’s changing.

In practical terms, System 2 thinking is what you use when you need to think deeply or critically about something. It’s the kind of thinking that requires you to stop and focus, rather than react on instinct or intuition. For example, when you’re learning a new skill, like playing a musical instrument or speaking a foreign language, you’re primarily using System 2 thinking.

Over time, as you become more proficient, some aspects of these skills may become more automatic and shift to System 1 processing. Understanding the distinction between these two systems is crucial in various fields, including decision-making, behavioral economics, and education, as it helps explain why people make certain choices and how they can be influenced or trained to make better ones.

AI System 2 thinking

Researchers are now striving to imbue AI with System 2 thinking to enable deeper reasoning and more reliable outcomes. The current generation of LLMs can sometimes produce answers that seem correct on the surface but lack a solid foundation of analysis. To address this, new methods are being developed. One such technique is prompt engineering, which nudges LLMs to unpack their thought process step by step. This is evident in the “Chain of Thought” prompting approach. Even more advanced strategies, like “Self-Consistency with Chain of Thought” (SCCT) and “Tree of Thought” (ToT), are being explored to sharpen the logical prowess of these AI models.

The concept of collaboration is also being examined as a way to enhance the problem-solving abilities of LLMs. By constructing systems where multiple AI agents work in concert, we can create a collective System 2 thinking model. These agents, when working together, have the potential to outperform a solitary AI in solving complex issues. This, however, introduces new challenges, such as ensuring the AI agents can communicate and collaborate effectively without human intervention.

Other articles you may find of interest on the subject of training large language models :

To facilitate the development of these collaborative AI systems, tools like Autogen Studio are emerging. They offer a user-friendly environment for researchers and developers to experiment with AI teamwork. For example, a problem that might have been too challenging for GPT-4 alone could potentially be resolved with the assistance of these communicative agents, leading to solutions that are not only precise but also logically sound.

What will AI be able to accomplish with System 2 thinking?

As we look to the future, we anticipate the arrival of next-generation LLMs, such as the much-anticipated GPT-5. These models are expected to possess even more advanced reasoning skills and a deeper integration of System 2 thinking. Such progress is likely to significantly improve AI’s performance in scenarios that require complex problem-solving.

The concept of System 2 thinking, as applied to AI and large language models (LLMs), involves the development of AI systems that can engage in more deliberate, logical, and reasoned processing, akin to human System 2 thinking. This advancement would represent a significant leap in AI capabilities, moving beyond quick, pattern-based responses to more thoughtful, analytical problem-solving. Here’s what such an advancement could entail:

  • Enhanced Reasoning and Problem Solving: AI with System 2 capabilities would be better at logical reasoning, understanding complex concepts, and solving problems that require careful thought and consideration. This could include anything from advanced mathematical problem-solving to more nuanced ethical reasoning.
  • Improved Understanding of Context and Nuance: Current LLMs can struggle with understanding context and nuance, especially in complex or ambiguous situations. System 2 thinking would enable AI to better grasp the subtleties of human language and the complexities of real-world scenarios.
  • Reduced Bias and Error: While System 1 thinking is fast, it’s also more prone to biases and errors. By incorporating System 2 thinking, AI systems could potentially reduce these biases, leading to more fair and accurate outcomes.
  • Better Decision Making: In fields like business or medicine, where decisions often have significant consequences, AI with System 2 thinking could analyze vast amounts of data, weigh different options, and suggest decisions based on logical reasoning and evidence.
  • Enhanced Learning and Adaptation: System 2 thinking in AI could lead to improved learning capabilities, allowing AI to not just learn from data, but to understand and apply abstract concepts, principles, and strategies in various situations.
  • More Effective Human-AI Collaboration: With System 2 thinking, AI could better understand and anticipate human needs and behaviors, leading to more effective and intuitive human-AI interactions and collaborations.

It’s important to note that achieving true System 2 thinking in AI is a significant challenge. It requires advancements in AI’s ability to not just process information, but to understand and reason about it in a deeply contextual and nuanced way. This involves not only improvements in algorithmic approaches and computational power but also a better understanding of human cognition and reasoning processes. As of now, AI, including advanced LLMs, primarily operates in a way that’s more akin to human System 1 thinking, relying on pattern recognition and rapid response generation rather than deep, logical reasoning.

The journey toward integrating System 2 thinking into LLMs marks a pivotal moment in the evolution of AI. While there are hurdles to overcome, the research and development efforts in this field are laying the groundwork for more sophisticated and dependable AI solutions. The ongoing dialogue about these methods invites further investigation and debate on the most effective ways to advance System 2 thinking within artificial intelligence.

Filed Under: Technology News, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

Let AI fully control your PC to complete tasks autonomously

Self-Operating Computer Framework let's AI control your PC autonomously

The idea of a computer that is capable of thinking and acting on its own is no longer a distant dream. thanks to this unique demonstration created using ChatGPT vision. Artificial intelligence (AI) has brought us to the brink of a new era where machines can perform tasks without human intervention.

Self-Operating Computer Framework – A framework to enable multimodal models to operate a computer. Using the same inputs and outputs of a human operator, the AI model views the screen using ChatGPT Vision and decides on a series of mouse and keyboard actions to reach an objective.

It is worth mentioning that at the current time GPT-4 Visions error rate in estimating XY mouse click locations is currently quite high. However this framework aims to track the progress of multimodal models over time, aspiring to achieve human-level performance in computer operation.

Control your PC using AI

This fascinating development is not just for tech experts; it’s something that anyone with a bit of technical know-how can explore and even set up themselves. Thanks to the recent rollout by OpenAI of its new ChatGPT creation service that enables anyone to create customized AI models without writing a single line of code in just a few minutes. For more information on how to create custom GPT AI models jump over to our previous article

Here are some other articles you may find of interest on the subject of AI automation :

Self-Operating Computer Framework

At the heart of this demonstration is AI, that is been used to enables machines to mimic human-like thinking. AI systems are designed to process visual data, make sense of complex information, and take action all by themselves. When integrated into a computer, AI transforms it into an independent operator, capable of starting up applications and browsing the web without a person’s input.

If you’re intrigued and want to experience this firsthand, you can create your own AI-driven computer. Begin by visiting GitHub, a platform where developers share their work. Look for a project by Other Side AI and use it as your starting point. The next step is to set up a Python environment on your computer. Python is a popular programming language in AI development because it’s powerful yet approachable. Once you’ve got Python up and running, activate it and install the necessary components for the AI to function.

  • Compatibility: Designed for various multimodal models.
  • Integration: Currently integrated with GPT-4v as the default model.
  • Future Plans: Support for additional models.this

Before the AI can take the reins, you’ll need to tweak some settings. Adjust the environment variables so your computer knows where to find the AI’s files. Then, change your system’s permissions to allow the AI to interact with your operating system. This lets it perform tasks like opening files and running other software.

Now comes the exciting part: watching the AI in action. You’ll see it navigate your computer’s interface, recognizing icons and menus by sight. It can simulate mouse clicks and keystrokes to open applications and browse the internet. The AI’s ability to search online is particularly impressive, showing its skill in finding and processing web-based information.

The benefits of letting AI have complete control of your PC

  • Automation of Repetitive Tasks: AI can automate tasks that are repetitive or routine. For instance, it could manage email sorting, automate data entry, or handle file organization based on visual cues and learned patterns.
  • Enhanced Accessibility: For individuals with disabilities, an AI with screen-reading capabilities could greatly improve computer accessibility. It could interpret visual information and convey it in alternative formats, like audio or simplified visuals, aiding users with visual impairments.
  • Efficient Troubleshooting and Support: In IT support and troubleshooting, AI could visually identify issues on the screen, guide users through fixes, or even resolve problems autonomously, thereby improving efficiency and reducing downtime.
  • Learning and Adaptation: An AI system can learn from the user’s behavior, preferences, and frequent tasks. Over time, it could adapt to optimize workflows, suggest shortcuts, or reorganize interfaces to suit the user’s habits.
  • Real-Time Translation and Assistance: For users interacting with content in foreign languages, the AI could provide real-time translation. It could also offer context-sensitive help in applications, improving user experience and productivity.
  • Enhanced Security and Monitoring: With the ability to continuously monitor the screen, AI could detect suspicious activities, like phishing attempts or unauthorized access, and alert users or take preventive actions.
  • Integration with Other AI Services: The AI could interface with other AI tools like language models, predictive analytics, and more, providing a seamless integration of various AI capabilities for a more comprehensive user experience.

Privacy and security concerns

Granting an AI system full control of a computer, combined with the ability to visually interpret the screen, can bring several benefits, especially in areas requiring automation, accessibility, and enhanced user interaction:

While the capabilities of AI-driven computers are exciting, they also raise important questions. What does it mean to give AI this level of control? Are there risks to our security or privacy? As you explore what your self-operating computer can do, it’s important to consider these issues. Understanding both the power and the potential risks of AI is essential as we begin to incorporate these systems into our daily lives.

The rise of self-operating computer systems steered by AI is a significant milestone in tech innovation. By following the steps outlined, you can set up your own system and discover its capabilities. As AI technology continues to advance, it’s crucial to stay informed about its impact, ensuring that we integrate it into our lives thoughtfully and responsibly. For Quickstart instructions jump over to the official GitHub repository.

Filed Under: Guides, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.

Categories
News

How to fine-tune ChatGPT 3.5 Turbo AI models for different tasks

How to fine-tune ChatGPT Turbo

We have already covered how you can automate the fine tuning process of OpenAI’s ChatGPT 3.5 Turbo but what if you would like to fine tune it for a specific task. AI enthusiast and YouTuber All About AI has created a great instructional video on how to do just that. Providing insight on how you can use the powerful ChatGPT 3.5 Turbo AI model to accomplish a wide variety of different tasks, training using specific data.

The process of fine-tuning the ChatGPT 3.5 Turbo model for a specific task, which in this case is to generate responses in CSV format compares the performance of ChatGPT 3.5 Turbo with GPT-4. When it comes to fine-tuning an AI model like ChatGPT 3.5 Turbo, the goal is to enhance its ability to handle the nuances of a particular task. By focusing on this fine-tuning, you can significantly improve the model’s ability to generate structured outputs, such as CSV files, with greater accuracy and relevance to the task at hand.

The foundation of any successful fine-tuning effort is a high-quality dataset. The adage “garbage in, garbage out” holds true in the realm of AI. It’s crucial to ensure that the synthetic datasets you create, possibly with the help of GPT-4, are varied and unbiased. This is a critical step for the model to learn effectively.

When comparing ChatGPT 3.5 Turbo with GPT-4, you’re looking at two of the most advanced AI language models available. Their performance can vary based on the specific task. For tasks that involve generating structured CSV responses, it’s important to determine which model can be fine-tuned more effectively to produce accurate and reliable outputs. GPT-4 boasts advanced capabilities that can be utilized to generate synthetic datasets for fine-tuning purposes. Its ability to create complex datasets that mimic real-world scenarios is essential for preparing the model for fine-tuning.

Fine tuning ChatGPT 3.5 Turbo

Here are some other articles you may find of interest on the subject of fine tuning large language models :

Once you have your synthetic dataset, the next step is to carefully select the best examples from it. These examples will teach the AI model to recognize the correct patterns and generate appropriate responses. It’s important to find the right mix of diversity and quality in these examples.

To start the fine-tuning process, you’ll use scripts to automate the data upload. These scripts are crucial for ensuring efficiency and accuracy when transferring data to the AI model. With the data in place, you can begin fine-tuning. After fine-tuning, it’s necessary to understand the results. This is where performance metrics come into play. They provide objective evaluations of the model’s accuracy, responsiveness, and reliability. These metrics will show you how well the model is performing and whether it needs further refinement.

The last step is to thoroughly test the fine-tuned ChatGPT 3.5 Turbo model. It’s essential to confirm that the model can reliably handle the task of generating structured CSV responses in a variety of scenarios. Fine-tuning AI models like ChatGPT 3.5 Turbo opens up a wide range of possibilities for tasks that require structured outputs. Whether it’s generating reports, summarizing data, or creating data feeds, the potential applications are vast and varied.

Refining ChatGPT 3.5 Turbo for CSV response generation is a detailed process that requires careful planning, the use of high-quality datasets, and a thorough understanding of performance metrics. By following the steps outlined in this guide, you can enhance the model’s capabilities and tailor it to your specific needs, ensuring that the AI’s output is not just insightful but also well-structured and actionable.

Filed Under: Guides, Top News





Latest timeswonderful Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, timeswonderful may earn an affiliate commission. Learn about our Disclosure Policy.