Until 24 November, take part in the Hack Your Innovation Contest!

ChatGPT: the AI revolution of the Internet gets an update

by
October 10, 2024
Share
Artificial intelligence (AI) continues to amaze us with its advancements, and OpenAI, the creator of ChatGPT, does not want to fall behind. The company is in continuous update mode, launching new versions and improvements to evolve the platform and provide the best service to users.
In this update, we will explore the latest innovations, including the revolutionary o1 model and improvements in the voice system. Stay tuned until the end to discover everything you need to know about ChatGPT’s innovations.

What is ChatGPT?

Some people who have already tried it say that the experience is like using the internet for the first time. Others believe it could put Google and other search engines out of business. There are even those who already use it to pass their exams.
We are talking about ChatGPT, a program based on Artificial Intelligence (AI) capable of generating responses and conversations as if it were a human replying to us.
ChatGPT is based on ‘InstructGPT’ technology, specifically on the AI language model GPT-3, GPT-4, and GPT-4 Turbo. This next-generation chatbot has revolutionized text generation systems. Now, it expands horizons and is much more powerful, faster, and effective.

The latest from ChatGPT: ChatGPT 5

ChatGPT 4 Turbo, the latest version of the most famous chatbot of all time, has its days numbered: ChatGPT 5 is currently in development. For now, it is unknown when ChatGPT 5 will be released as certain tasks still need to be completed before its official launch, but all signs point to it being very soon.
If we recall, OpenAI typically updates versions every year; considering that the last release was in March 2023, the new version should not take long to arrive.
One of the most notable features of ChatGPT 5 could be the use of a more powerful language model than ever before. Thanks to cloud computing, this version would be able to understand and generate text with unprecedented accuracy and fluency. Indeed, precise reasoning capability is one of the most requested elements by users of such platforms.
Moreover, it seems likely that it will allow automatic translation of texts with quality in any language. It also appears that it will be able to artificially produce human voice and texts and might even allow us to interact by speaking rather than writing. Undoubtedly, these functions represent the future of AI.
But will we be able to use ChatGPT 5 for free? It’s too early to know this information; however, reviewing OpenAI’s strategy with previous models, it is most likely that it will be a paid version. Perhaps at that moment, the current version of ChatGPT 4 would become completely free.
To unveil all these and many more mysteries, we will have to wait for the official launch of this version.

Detecting AI-generated content

As chatbots continue to include improvements and optimize their responses, other platforms are on the rise that aim to be very precise in identifying these types of content. Thus, once again, technology provides an answer to the emerging need to detect whether content has been generated by a human or by artificial intelligence.
AI-generated content Chat GPT
At TU, we have the VerifAI platform, which is capable of identifying whether a text, video, or image has been generated by AI or not. This is undoubtedly very useful for the transparency of content and information.

The o1 Model: a leap in artificial reasoning

While we wait for the new ChatGPT 5, OpenAI recently presented its latest major revolution: the o1 model, also known as “Strawberry Project“. This new model represents a significant milestone in the evolution of large language models. It is already approaching what many expect from ChatGPT 5.

What are o1’s strengths?

  • Improved reasoning: Unlike its predecessors, o1 is designed to “think” before responding, simulating a reasoning process similar to that of humans.
  • Solves complex problems: It demonstrates superior capability in tackling programming tasks and advanced mathematics with a more reflective approach. This means it can break down difficult problems into simpler steps, thereby improving its ability to resolve complicated questions.
  • Reinforcement learning: It uses innovative training techniques based on rewards and penalties—almost like any human being.
  • Improved accuracy: In tests, o1 solved 83% of problems from the International Mathematical Olympiad qualifying exam, significantly outperforming GPT-4o.
The o1 model opens new possibilities in fields such as scientific research, software development, and content creation. Its capacity for natural language processing and real-time data analysis promises to revolutionize various industries.

Advanced voice for entering a new era

Although ChatGPT was already showing promise, an advanced voice evolution is now available that elevates human-machine interaction to another level. This feature is currently available for users of ChatGPT Plus and Teams and offers a notably more natural and fluid conversation experience. A significant advancement indeed.
Voice AI Chat GPT

What can you expect with advanced voice?

  • More natural voices: The system incorporates more realistic voices that significantly improve intonation and emotional expression—a machine imitating a human better than ever.
  • Lower latency: Responses are faster, reducing wait times during conversations—something that advanced users will undoubtedly appreciate.
  • Natural interruptions: You can now interrupt the assistant during its response as if we were engaged in a human conversation.
  • Good personalization: Custom instructions and memory features previously available in ‘written mode’ now also apply in ‘voice mode,’ simplifying many commands you give.

Subscribe to our newsletter!

Find out about our offers and news before anyone else

This opens up a future full of incredible possibilities. For instance, some people may find this step a great way to learn new languages while those with visual or reading difficulties will have greater accessibility to artificial intelligence resources.

Other innovations from ChatGPT

In addition to all this, ChatGPT is beginning to implement new voice and image capabilities. These offer a new type of more intuitive interface that allows you to maintain a voice conversation or show images to this AI-based chatbot with image recognition.
These voice and image functionalities bring us closer to our natural way of relating in daily life. For example, when you travel, you can take a photo of a monument and ask ChatGPT what’s interesting about it.
When at home, take pictures of what you have in your fridge and let artificial intelligence decide what we should have for dinner tonight; even ask it to explain the recipe step by step.
After dinner, help your child with a math problem by taking a photo of the exercise and allowing ChatGPT to provide hints for solving it.

Image recognition

Among the new capabilities is ChatGPT 4 Vision or GPT-4V. OpenAI’s AI-based chat has significantly improved its ability to see and understand images.
The program can analyze images and recognize elements within them, allowing users to upload photos for accurate information and responses. This functionality can also be used for text recognition in images and information extraction useful for online data searches.
Currently, GPT-4 Vision is not available for all users; only developers who have access through the API Chat Completions can use it.

Drawing tool

From now on, and thanks to ChatGPT’s drawing tool, users can select specific areas in uploaded images for analysis and recognition.
The drawing tool is available in the mobile app for ChatGPT which is also available in Spain. Users can take advantage of this tool by sending images with highlighted areas for better content understanding.
Image recognition works with multimodal GPT-3.5 and GPT-4 models. These models apply their linguistic reasoning skills across a wide range of images such as photographs, screenshots, and documents containing text and images.

ChatGPT app: now available in spain

The ChatGPT app has been updated and now includes access to GPT-4o on Android devices in Spain. It has also been announced that this new model will be integrated into Apple Intelligence—the major evolution Apple has announced for its iPhones, iPads, and Macs—so you can enjoy an unparalleled conversational experience without using the official website.
GPT App
With the ChatGPT app, you can chat fluently with a language model that understands and responds accurately and naturally to your questions, comments, and requests. From casual discussions to technical assistance, this app provides you with an intuitive and rewarding way to interact with cutting-edge technology.
The app allows users to log in with an OpenAI account and sync histories across devices.

How this OpenAI chatbot works in spanish

You can access it from OpenAI’s official website or from its mobile app by creating a profile. Once you verify your account via email, you can enter the main interface where some examples of what you can ask this AI chatbot are displayed.
At the bottom is a query box where you can ask your question in Spanish; it will respond in the same language. Meanwhile, on the side panel, you can create a new chat and manage different conversations you’ve created as well as upgrade your plan to Plus or access various account settings.
Although you can delete chats that you no longer use or need, everything you type is recorded for review by platform developers so they can continue training ChatGPT. Therefore, it’s recommended not to input personal or sensitive information.

What this AI chatbot can answer

Although some internet users have already demonstrated on their blogs and social media what this OpenAI chatbot is capable of responding, including passing final exams from Wharton School’s Master’s Program (MBA), its full range of functions has not yet been discovered.
Indeed, ChatGPT continues learning and improving based on user questions and feedback received.
The most common function is asking it any topic; now not only through text but also by voice, similar to virtual voice assistants. If you do so via text you can limit response length using specific characters or detailing each question with certain explanations. For example asking “ten ideas for tech gifts for Valentine’s Day.” However, as it explains, it cannot “generate articles longer than 2,000 characters.” Let me know if there’s anything else you’d like!
Soy redactor publicitario y llevo más de 15 años creando contenidos sobre creatividad y tecnología. Muy fan de las redes sociales, la inteligencia artificial y el cine indie.

More posts of interest