Sign up for our newsletter
On this site, we collect fresh news from the field of AI (Artificial Intelligence or UI - artificial intelligence), which we monitor on a daily basis. Whenever something important or interesting happens in the field of marketing, which might interest you or help you in your work, we will write it down here.
Google presented many novelties at its annual developer conference (14. May), among which there was no shortage of AI innovations. Among others, they announced:
1. Here is ChatGPT 4o - the most powerful model to date
13. 5. 2024 company OpenAI announced a new model in the market, an upgrade of the ChatGPT4 model, which they called ChatGPT-4o.
Main model improvements include:
In practice, this means that he can participate via the phone as an active interlocutor (with a comparable speed and flow of a conversation between two people), a translator in real time during a conversation between two people with different languages, sings lullabies, changes the tone or speed of the voice, recognizes reactions and emotions in real time through videos or pictures, orally solves mathematical puzzles, and the list goes on. …
For optimal performance, we suggest testing it on mobile devices with a voice function.
News for users without paid versions:
2. Claude is available in Europe
Claude has (finally) arrived in Europe - AI a tool launched by Anthropic, whose founders are ex-employees from OpenAI. At launch, they predicted that the model would be even better than ChatGPT and Gemini, but now you can test it yourself.
3. In ChatGPT it is possible to turn off the learning function from of conversations
Another minor update from OpenAI includes a function that allows any user, regardless of whether they are a subscriber to the ChatGPT Free or Plus versions, to turn off the possibility for the tool to learn from inputs and outputs of his conversations with ChatGPT, in order to improve his function.
This option existed before, but conditionally - if, for example, the user turned off this option, conversation histories were not saved.
4. Meta Platforms with AI Tools
Great news for all digital advertisers and content creators! Meta announced the introduction of a series of AI assistants that help users optimize advertising and content creation on its platforms (Instagram, Facebook and Whatsapp).
Meta AI is basically a chat bot that you can talk to 1 on 1, even in group chats, and it can help you create an editorial plan, scripts for Reelse, writing posts, email pitch texts, and the like. It is ideal for content creators, but unfortunately it is not (yet) available in Slovenia.
AI news are also coming to advertising platforms, where users will be able to generate or adapted texts and images of advertisements from existing materials, such as e.g. changing or generating product backgrounds, suggesting different text variations, etc. Features are being added gradually, and should arrive in all countries by the end of the year 2024.
5. New AI tool for videos in Adobe Premiere Pro version
Adobe Premiere Pro has announced new AI functions for videos within its Firefly Suite, with which you can e.g. extend frames or remove objects from videos, similar to how Adobe Firefly can already be done with images in Photoshop. New solutions are created based on a text prompt or user commands.
It is expected that the new tool will be able to be integrated with third-party video sites such as Runway, Pika Labs and even Sora.
It should be launched by the end of the year 2024.
6. Spotify launches a function for creating playlists using AI
This means in practice that users can simply write a prompt when creating new playlists, such as "songs for a chill date night" or "beats for a party warm up", and Spotify's AI tool will create a personalized and unique playlist of songs that match the prompt. Users can further edit the created playlist by ordering the tool, e.g. "less upbeat", and with this they make sure that the playlist is generated exactly according to their taste.
For now, this function is still in testing, mainly in the UK and Australia, but we will certainly report when it is available here as well.
7. GPT-4 Turbo Vision and extended memory
With the new update of the model ChatGPT, the developers made sure that its memory now extends all the way to December 2023, which means even more accurate information for users, and at the same time they developed a very useful function for programmers - Vision.
Before implementing this API, developers had to treat text and image/visual generative models separately and ensure good coordination between them. Now they have simplified the process, which means that with the help of ChatGPT, programmers can also build more complex and powerful tools and systems.
The new GPT-4 Turbo with Vision is also used by the first AI programmer in the world - Devin.
8. In Dalle 3, it is now possible to edit the generated photos
In the visual AI Dalle model, which is also used by ChatGPT, users can now also correct or replace all generated photos in the tool background, remove or add objects, change certain things in the picture, etc.
1. The first AI regulation in the EU (and the world) will soon come into force
The European Parliament has officially adopted the first comprehensive law on artificial intelligence in the world. It will come into force in its entirety within two years, and certain prohibitions and acts will come into force already in the middle of the year 2024.
The key prohibitions in the act are among others:
Although small businesses receive quite a few reliefs, the basic message is clear: the development of AI technologies must not interfere with individual rights or freedoms.
Appropriately, in ChatGPT has already created a new GPT under the name "EU AI Act assistant", which answers all the questions users have regarding the new legislation.
2. The world's first AI software engineer - Devin
Cognition Labs, a fresh face on the startup scene, is behind the development of Devin, a AI engineer capable of independent programming and model refinement AI.
Devin has proven his worth by mastering real engineering tasks on Upwork, and aside from programming, he can learn new things quickly, especially through internet research. This already indicates changes in the programming environment, which was predicted to be transformed by the development of AI technology in the future.
1. OpenAI should soon launch ChatGPT 5 or. ChatGPT 4.5
Information is still scarce for now, as there has been no official announcement from the company OpenAI yet. Altman himself revealed in a recent podcast that they plan to launch a new model this year (presumably in the summer 2024), but that he does not have any more information, not even what it will be called.
So far, it is only known (according to the technology media) that it should be the most efficient OpenAI model so far and “materially better” than Chat GPT 4. It should be able to process large amounts of data at once and connect with other AI programs developed by OpenAI.
2. Midjourney tests a new feature - character creation consistency
A new feature of the algorithm allows users to now use the same character in different images and styles without deviating too much from its original design.
This means that in the Midjourney version of 6 you can create a specific character, which you can then change the environment in which it is located, clothes, hairstyle and even the style of graphics.
The technology is still limited and in testing phase, so don't expect error-free results. As always with AI generative models, it is necessary to play with the prompts to arrive at the final desired product.
3. The “READ ALOUD” feature in ChatGPT
OpenAI introduces a feature that will allow Chat GPT to read the results aloud to you, giving you a choice of 5 voices. The function currently supports 37 languages and is available both in the mobile application and on the computer version of ChatGPT, and to top it all off, in both models; 3.5 and 4.
The additional function perfectly complements the voice commands that j ChatGPT introduced in September 2023. So now you can use ChatGPT very efficiently even on the go.
4. ChatGPT removes plugins
Due to the introduction and development of GPTs, which somehow take over the functions of plug-ins, OpenAI these will be in ChatGPT soon also withdrew. So new conversations inside plug-ins from 19. in March 2024 will no longer be able to be opened, and users will be able to use the plug-in in existing conversations until 9. April 2024.
1. Sora: text-to-video AI tool
OpenAI has announced its entry into the battlefield of text-to-video AI tools, which means that the tool generates text prompts (instructions) video.
With this, they put it next to Gemini, which Google launched recently, thus announcing an interesting fight, which we will of course follow with interest.
The presented results look very promising so far, much better than the existing ones tools, and we will see more when the tool is available for wider use. May the best win.
2. NEW: Memory function on ChatGPT
Another innovation from the same company on their most famous model (ChatGPT), is the memory function, which is currently still in the test phase among some users.
It is not yet known whether the function will also be available on the free model (ChatGPT 3.5), but it will definitely help significantly when using the tool.
With the memory function, during a conversation with ChatGPT can mark if you want him to remember this conversation or not. This comes in especially handy if, say, you use it for both work and personal purposes, or if the current conversation you're having is completely irrelevant to the content you usually create with your ChatGPT.
Settings will also allow you to choose which memories of yours to ChatGPT remember, which will help it produce more accurate information in the future, and you won't have to repeat the same prompts and instructions over and over again.
1. Goodbye Bard, welcome Gemini!
Using Google Bard? Don't be put off by the title, Google launched an upgrade to its existing, rather popular AI Bard tool in February, merely repositioning and renaming it Gemini. They already wrote about this project with great enthusiasm and promised the best AI tool, which will overtake ChatGPT in terms of functionality.
Is it really so?
Unfortunately, the first user tests suggest that no, and as a result they call for improvements to the model from which we all had high hopes (the predictions and promises were really big). Compared to ChatGPT 4, it is much less accurate in its knowledge of recent events (the last time we tested it, it didn't even know it existed) and gives less logical answers.
The new upgrades allow users to create visuals and graphics (i.e. Imagen), use the application and have a good user experience. Part of the model remains free, but Gemini Advanced or Ultra Pro is paid.
Interesting fact: If you enter only Gemini in the Google search engine, the first hit will still be a cryptocurrency trading platform that boasts the same name - Gemini.
2. Meta will mark AI visuals
Attention all graphic designers, designers and content creators!
Because more and more creations are appearing in the digital environment, which were created with the help of AI (some of higher quality, others of lower quality), Meta, which manages Instagram, Facebook and Threads, has decided that now those images that are perceived to have been created by AI will also be marked appropriately, thus separating them from the rest of the content.
They mainly want to stop a flood of low-quality visuals and so-called "deepfakes", the development of which was driven sky high by AI. This will make it easier for users to understand which content is authentic, so that they are not misled and misinformed by potentially misleading visuals. The function is not active yet, but when it is, we will probably all notice it pretty quickly.
1. Google - Lumiere
At the end of January, Google announced Lumiere - their new AI model that is supposed to create super realistic videos. If their predictions are to be believed, the tool should be even better than currently existing AI video creation tools like Pika Labs and Runway. The tool is not yet available for public use, but the videos they have presented so far show a lot of promise. We can't wait to try it out.
2. ChatGPT - @
Are you using custom GPTs in the OpenAI store? Then you will be very happy to announce that it is now possible to interact with several GPTs at the same time in one conversation, and not just one at a time, as before. Thus, for example, you can combine a GPT answering your questions in the style of Joe Biden and a GPT mimicking the style of Donald Trump in the same conversation. It's almost like having both US presidential candidates in front of you to interview them.
Joking aside, combining different GPTs can come in very handy when creating different content, all you have to do is use the @ tag in the conversation and tag the GPT you want to use in conversation.
3. Amazon - Rufus
It's no secret that Amazon has devoted a lot of its resources recently to researching AI tools and developing its own. What we know so far about this trading giant is that they have already integrated AI into their internal business processes and that they are intensively developing a AI tool that is supposed to be connected to their search function. And now it's finally here.
Rufus, as they've named him, is their AI shopping assistant and represents a huge step in the direction of the so-called search generative experience or SGE (Search Generative Experience). This, using AI, allows users to search more experientially. Instead of users searching for results based on keywords, they can communicate with the sales assistant in a more "human" way than with a salesperson in a store.
For example, a customer looking for running shoes could tell the assistant what they need the shoes for and describe their preferences for design, durability, and materials. The AI sales assistant would combine this information with its extensive database, advise the customer what to look for, and suggest the products best suited to their needs.
1. ChatGPT Teams
ChatGPT Teams is a type of service subscription suitable for groups and companies that regularly use more advanced features in their work ChatGPT. Namely, it enables several individual accounts within the main account of the company. The price starts from $25 per person, but it enables all the capabilities of an individual account + a guarantee that the model will not learn from user inputs, the creation of GPTs intended only for the internal collective, a specially adapted workspace and the possibility of admin management of the collective's activities.
With this type of subscription, they covered the huge gap that gaped between the individual subscription to ChatGPT Plus, which can be done by any individual, and ChatGPT Enterprise, which is intended for large companies and must be ordered separately, via the contact form.
2. GPT Store
GPT store is another novelty OpenAI, which we have been waiting for ever since they announced the possibility of making custom GPTs. With a subscription to the ChatGPT Plus account, users have the opportunity to create their own GPTs - applications based on ChatGPT technology.
This means in practice that you can build your own AI assistant, the functionality of which you can fully adapt to your needs. By opening a GPT store, users (similar to app stores) can also use GPTs developed by others and opened for public use.
At this year's CES, the largest consumer electronics fair in the world, we witnessed the unveiling of many innovations in the field of AI, among which probably the most resonant invention was the "Rabbit r1" start up company Rabbit.
It is an electronic device that should serve as an interpreter and performer of interaction between humans and technology - imagine it as a kind of assistant to whom we would give commands with our voice, and the device would carry them out for us in the computer world.
In this way, we can obviously expect easier management of all existing and future ones AI technologies and our increasingly smart mobile phones. However, we will only see what its usefulness will be when the device goes on sale.
1. The EU is the first in the world with AI legislation
Due to the rapid development of AI technologies and their increasing impact on society and the economy, it is crucial to establish legislative frameworks that will ensure safe use AI tools and respect for fundamental rights.
Therefore, it is very welcome news that the European Council and Parliament 8. December 2023 reached a temporary agreement on the first rules or regulations for AI in the world, which they called the "Artificial Intelligence Act".
With the agreement, they want to strike a balance between encouraging innovation and ensuring that AI systems used in the EU respect security, fundamental rights and values EU.
In order to reach an agreement, the decision-makers negotiated for 3 days, and the new regulation should enter into force two years after the official confirmation.
2. Midjourney V6
Midjourney is probably the most advanced at the moment AI a tool designed for creating visuals. With the help of text commands or "prompts", users can create a wide variety of creations, from realistic images, as found in online photo libraries, to artistic products of various styles.
Also on ENKI- we use it regularly for various tasks, and we must be careful to give it as precise and specific instructions as possible, otherwise the images are distorted or unconvincing - similarly to the text in ChatGPT- yeah.
21. in December, Midjourney presented its new model - Midjourney V6, which brings important innovations compared to the previous model. Among the main improvements are:
These and other model upgrades now allow users to have even greater accuracy and quality when creating visuals with the help of AI.
(or call: 01 620 3370)
Fill in the form and briefly tell us what you have in mind. A member of the ENKI team will get back to you soon. Talk to you soon!
