Surely by now, you’ve heard of ChatGPT. Or DALL-E. These new technologies that are dominating conversations and debates between professionals and the general public are neither more nor less than artificial intelligence (AI). Specifically, we are looking at two examples of generative artificial intelligence. One more evolution within the field of AI with a very promising future.
Generative artificial intelligence (GAI) is based on deep machine learning methods. Basically, it is a form of artificial intelligence that takes machine learning one step further. Its operation involves collecting information about certain elements that will later be used by the machine to generate other ideas.
Basically, the algorithms of generative artificial intelligence create content from previously provided data. Data can be, for example, texts, images, videos, or music. Its creation from all this information—and without the need for human intervention—is, in theory, indistinguishable from what a person would make. It is, therefore, something revolutionary in any field and everything indicates that it will have an impact on the activity of companies, including their marketing and advertising departments.
Generative artificial intelligence is one of the most prominent strategic trends today. In fact, a study by the consulting firm Gartner predicts that by 2025 this type of artificial intelligence will represent 10% of all data produced. Likewise, IAB in its Top Digital Trends 2023 includes in the data section the emergence of AI in the world of marketing. And, in addition, it specifically mentions generative artificial intelligence:
“In 2023, artificial intelligence will be an element to take into account in marketing campaigns as it helps personalize customer experiences by increasing the effectiveness and efficiency of campaigns. An example of the above is the emergence of tools such as ChatGPT (based on the “RLHF” learning method), which are beginning to gain popularity and which, without being specifically designed for digital advertising, will be able to help media, search engines and advertisers to improve and energize its content and advertising”
Apart from digital advertising, there are other examples of generative artificial intelligence such as those pointed out by IEBS School, which range from the creation of personalized therapy bots to the explanation of scientific concepts or the writing of university essays.
How does generative artificial intelligence work?
As we have seen, the operation of generative artificial intelligence aims to be comparable to that of human intelligence. To do this, they use what is known as generative neural networks.
As explained by OBS Business School, these networks use deep learning to learn and analyze data and, in turn, find patterns that would be very difficult to find otherwise. A process that, gives rise to generative artificial intelligence, is completed with the use of GANs or generative adversarial networks.
This technology, which is somewhat more specific, allows unsupervised learning to be developed. They do this thanks to the two parts that make them up: the generator and the discriminator. The first of them is capable of creating new content and the second of analyzing whether it is real or false.
Likewise, within the technologies of generative artificial intelligence, we must also talk about GPTs or pre-trained generative transformers. This model, according to Cyberckick, uses statistical methods and is capable of understanding human language. Therefore, it is capable of creating completely new texts from scratch.
Uses (and risks) of generative artificial intelligence
As we have already mentioned, generative artificial intelligence has many uses, also in the business field. From helping in the creation of software code to 24-hour customer service, including R&D&i work.
There are Generative AI models with transformative capabilities in complex fields, such as computer engineering. For example, Copilot (GitHub – Microsoft) suggests code and helps developers automatically complete their programming tasks.
This technology applies to multiple industries. One of them is healthcare, a sector where thanks to it new medications can be developed or an x-ray of a patient can be compared with images of healthy organs to detect tumors.
Risks also accompany the IAG
The development of IAG, although it can be very useful, also brings with it some dangers. For example, applied to the case of ChatGPT, Cybersecurity PYME mentions the following risks of generative artificial intelligence :
- Doxing. This concept refers to the disclosure of private information, whether of a natural or legal person, over the Internet.
- Fake news. On the other hand, the misuse of generative artificial intelligence can lead to the creation and dissemination of false information.
- Pishing. The use of IAG can help cybercriminals refine the content of their scams and, consequently, they will be more difficult to identify.
- Deepfakes. In this case, images are created where the face of one person is superimposed on that of another to falsify their gestures and voice. On some occasions, deepfakes of famous people have been created that have managed to go viral.
- Identity fraud. Also, IAG poses a risk when it comes to protecting a person’s identity on the internet. This is because people seeking to impersonate or steal personal information will be able to rely on fake profiles that work with this technology.
The perception of these risks, which can be extended to other generative artificial intelligences, must serve to understand the scope of this technology to promote only its positive part and make ethical and responsible use of what will undoubtedly be, one of the innovations of the future.