Currently, there is hardly a hotter topic than that of artificial intelligence. Especially the name ChatGPT triggers something in many minds and is emblematic for the connection of AI with simple text inputs. ChatGPT-4 is the latest model of artificial intelligence and goes a big step further, in direct comparison with its predecessor ChatGPT-3.
But what innovations does ChatGPT-4 bring with it, and what should companies be prepared for now? Will AI be more of a help in the future or perhaps even a looming threat? How can companies benefit from this in the area of IT security, and how can targeted phishing attacks be prevented through ChatGPT?
We took a closer look at ChatGPT-4 as artificial intelligence. In doing so, we highlight opportunities and risks, go into the connections with Microsoft, tell you where ChatGPT-4 is in use and how you can use it, but also clarify the criticism of AI a bit. Because despite all the enthusiasm, technology should not be viewed only in a positive light.
What is ChatGPT?
The acronym ChatGPT stands for Generative Pre-trained Transformer. This is a chatbot trained using machine learning that outputs text-based answers as a result and can be used extensively. ChatGPT was developed by the US company OpenAI.
The basis for the chatbot is the language model GPT-3, which was published by OpenAI shortly afterward already in version 3.5 and now in version 4, i.e., as ChatGPT-4. The basis of GPT can be found in transformers from Google Brain. Based on this, OpenAI developed its own methods to train and significantly improve the chatbot.
The result is now version 4, ChatGTP-4, which has become even more complex and diverse, so that the answers should be even more creative. Especially the accuracy is something OpenAI currently has to pay special attention to, since ChatGPT-3 still mixed up many facts or simply invented new ones. So is ChatGPT-4 really the longed-for breakthrough in terms of artificial intelligence?
Why is ChatGPT trending right now?
ChatGPT is currently in vogue mainly because it is the first publicly understandable AI that really works in a meaningful way. Up to now, there have been attempts at artificial intelligence, but these were either only known to IT experts or were simply too complicated to be immediately understood, let alone used.
However, with ChatGPT from OpenAI, this point has now changed significantly. In principle, anyone can use AI as a chatbot, since we as a society are already used to chats and simple bots with automated responses no longer seem to be a rarity. For example, these often occur in support chats before a real employee responds to a request. So we already know about bots, and we talk to friends and family via chats in WhatsApp and iMessage. The AI acts like an artificial friend that has an answer to every question.
So ChatGPT is on everyone’s lips because it’s the first AI that anyone can just try and use. And the first one that really produces many, meaningful results, even if the pure quality is not always there. The AI should not be trusted unreservedly when texting. As a help, AI is great, but it is also just like that one friend who shares his half-knowledge with you before admitting he has no idea about the subject.
Where can ChatGPT-4 be used?
Microsoft has invested billions in OpenAI. It is therefore not surprising that the company was one of the first to be allowed to use ChatGPT-4 and also rolled it out accordingly on a large scale. Artificial intelligence has now found its way into almost all of Microsoft’s products, and the expansion is just beginning, it seems. For Microsoft, the topic of AI is currently essential.
While at the beginning it was ChatGPT-3 that was used by Microsoft, Bing can now be used with ChatGPT-4. The new Bing is available for Edge, it is integrated into Skype, and artificial intelligence is also a priority in Microsoft Office. It is remarkable how quickly Microsoft currently reacts, optimizes and brings ChatGPT-4 into its products. AI is increasingly being integrated into products, including NotionAI, for example.
The strategy that Microsoft is pursuing is relatively clear. Through the interest in artificial intelligence and specifically the capabilities of ChatGPT, the company’s own products are being advanced. First and foremost, the Bing search engine and the Edge browser. Both very popular, but behind Google and Google Chrome. However, ChatGPT has now massively increased interest in both products, as they are naturally required for use. So Microsoft is gaining market share through its AI deployment.
What improvements does ChatGPT-4 bring?
First of all, the model has simply become more powerful. It handles more complex queries better, can process up to 25,000 words at once, and is said to respond to queries much more creatively than it has in the past. This logically results in more coherent answers with ChatGPT-4, which make more sense and are often much more diverse than with ChatGPT-3.
OpenAI has probably also paid special attention to the interpretability of ChatGPT-4. This mainly affects images. ChatGPT-4 should not only be able to describe them excellently with words, but should also process images in general for its replies. An interesting and still new aspect. ChatGPT was recently very text-heavy and worked much less visually. Will ChatGPT-5 then maybe even have videos to watch?
The new method makes it possible with ChatGPT-4, among other things, to easily photograph tasks and have the artificial intelligence solve them. The new word count allows scientific studies to be uploaded and turned into short summaries. In short, ChatGPT-4 has become more comprehensive, even smarter and significantly more versatile. There are fewer limits in the data that ChatGPT-4 can now use for its own responses. Currently, anything seems possible as long as a text-based response is the desired result.
Criticism of OpenAI and ChatGPT-4
A major criticism is that OpenAI has massively changed the open and scientific strategy. With its success in Bing search, OpenAI has to some extent changed course and is now no longer pursuing an open approach. This probably happened after the great success and massive investment of Microsoft or its use in the search engine Bing.
For example, ChatGPT-4 does not have any documented details about the exact technology, hardware, or training data. Similar to Google’s algorithm and its rankings of websites, there is now no indication of how ChatGPT-4 filters or even presents content. This makes the AI a kind of black box that can no longer be looked into. How does it deal with racism, illegal requests or sexuality?
There is a lot more criticism apart from that. Regarding the partnership with Microsoft, misinformation from ChatGPT itself or even the lack of transparency in the background. The impact that such AI systems will have on everyday life, universities, schools, and daily living is also still uncertain and poses many social risks. Most recently, this ended with many calling for an AI development pause to address just such ethical and sociological issues.
What are the dangers of ChatGPT?
The AI development pause just mentioned is no accident. For one thing, there is presumably only one company right now, OpenAI, that can provide powerful AI. This already bears the danger of monopoly, especially if Microsoft as an investor is the sole major user of ChatGPT. Second, many ethical questions arise regarding artificial intelligence and its use in daily life.
What happens when homework or dissertations are done with ChatGPT? Will books, texts, and websites be written only by AI in the future and how valuable or artistic will they be then? Where does the AI stop if it starts generating images or mimicking styles now? Is it even allowed to do that? Will there soon still be “real” content, or will the majority of all content soon have been artificially created by an AI?
The AI development pause is also called for because many fundamental questions in the field of artificial intelligence have not yet been conclusively resolved. Thus, there are no security protocols, watermarks, liability schemes or certifications. Content that the AI learns from is often stolen because it has been illegally processed. However, no one can exclude their own content, as this has not been provided for so far. So ChatGPT learns from everyone without giving anything in return.
So what does artificial intelligence do socially? What kind of damage does it cause? It could easily find hacks itself in the future or quickly expose security vulnerabilities. Already, reports of hacks using ChatGPT continue to grow. How can IT infrastructures be protected when an AI is permanently looking for gaps? Is IT security then still possible at all?