Zvonimir Petković: By the end of the year, almost all of us will be using generative artificial intelligence.

If you talked to Zvonimir Petković, or were at his lecture, you probably left with knowledge that you could use immediately in your work. A connoisseur from whom many want to learn, Zvonimir is a software engineer at Infobip, under the umbrella of the EDIH Adria consortium Holds educations on how to improve business using generative artificial intelligence.

They are free for employees of small and medium-sized companies and public administration, associations and institutions in Adriatic Croatia. The next one is on March 20th in Infobip's Quantum in Vodnjan, modernly equipped space in Istria 9.

Undoubtedly, generative AI has been a major technology topic for more than a year since ChatGPT emerged at the end of 2022. It is projected to bring about a major re-allocation of jobs and affect many aspects of the economy and society. According to some forecasts, as a result of the development of genAI around 300 million jobs could disappear worldwide, and at least so many new jobs could be created. Although this technology is not omnipotent, and is not a magic wand that will solve all problems, it is wise to know as soon as possible what it is, and how to use it in everyday work.

GenAI: Where to start if you are a public authority or a small business?

Regardless of the size of the company or institution, and the area it deals with, there are several examples of the use of generative artificial intelligence that suit almost everyone, says Zvonimir.

-First, it is the removal of the burden of monotonous, repetitive tasks from employees, which in many cases can be achieved by implementing chatbots. Second, there is the production of text and images. Thirdly, Microsoft will introduce Copilot into all its applications by the end of the year, which is the name for their genAI solution. We will have a Copilot in Word, Excel and PowerPoint, so it is likely that by the end of 2024 almost everyone will use genAI, or at least be in the immediate opportunity to use it, explains Zvonimir, who has significant experience in both education and new technologies.

After studying at the Faculty of Mechanical Engineering and Naval Architecture in Zagreb, he began to work in teaching, both at school and at the faculty. He taught programming in Python, then 3D design, and several other fields of mechanical engineering at the Polytechnic of Technology and in high school.

-In the public sector, you often have your own job and a number of additional responsibilities. So at one time I was also a shift manager with a lot of administrative tasks. With a few other people, I installed a network infrastructure – several kilometres of cables. At some point I was responsible for the operation and maintenance of 300 computers, reveals Zvonimir.

In Infobip he specializes in the implementation of generative artificial intelligence, good practices and increasing the efficiency of internal processes through this technology.

One thing is certain: You won't get it from him. ex cathedra lecture. It quickly identifies the needs and opportunities for implementing genAI in companies and public authorities. It will guide you to useful services and in an interesting way shed light on the concepts and trends in generative artificial intelligence, as well as the risks that need to be taken into account.

Some believed that if they didn't speak the best English, they wouldn't be able to use genAI. However, this is not true. Others assumed that geneAI tools were complicated, but it took them less than 10 minutes to master them and recognize how to use them in the workplace.

Chatbot: Remove the burden of repetitive work from employees

Just five or seven years ago, chatbots were not developed enough to be entrusted with more complex tasks. Today the situation is different. Chatbots can automate exhausting communications that constantly repeat the same questions and answers. Instead of the employee answering the same questions by email or phone from day to day, this job can be transferred to a chatbot. This frees people time for other kinds of jobs, which bring more value to the organization.

If you call Zagrebačka banka, you will talk to virtual assistant Mia. LAQO Croatia insurance handles about 30 percent of customer inquiries via chatbots. Podravka's Coolinarika has also implemented a chatbot. Some smaller companies have done the same. It is also easy to imagine that a school, an association or a utility company, instead of giving information by phone or email, solves some repetitive questions via chatbots. Whether they implement it on their website, on WhatsApp channel, or in their call center, will depend on the possibilities and needs.

-You get the most benefit from a genAI chatbot when it has access to your company's or institution's data. Then the answers he gives to the user's questions are of the highest quality. Small businesses or smaller municipalities often do not have a technical employee who would do the integration themselves, and in most cases they hire an external partner for this job. On the other hand, medium-sized companies and institutions will often have at least one person among their employees who will be able to do data integration or conduct this business with external partners, Zvonimir explains.

Small healthcare, transport, construction and utilities companies

The path towards education on genAI and its implementation is not the same for all small and medium-sized enterprises. For example, a quality dentist or an intern with a private shop and two employees will hardly set aside five days for training on increasing competitiveness through new technologies because the pace and amount of work does not allow it. Still, as business owners, they think in the long run and know that new technologies will be an increasingly important factor in business. How do they access generative artificial intelligence?

-If you mention small dental practices, they can benefit from AI image generation, for example when they want to show the client their future condition, after dental implants have been installed. This image can be produced by a genAI based on a photo of an existing situation and a few simple instructions. Smaller transport, construction and other businesses, as well as smaller municipalities, can save time and speed up processes by using tools such as ChatGPT to help create memos, emails and logs. An excellent example is the use of genAI to help write applications for EU projects: give him all the project documentation and then ask him what you can apply, given the type and needs of the company. They will retrieve data and provide answers, which can be of great help in compiling documentation. Of course, you should always take care of data security and check the generated content well, clarifies Zvonimir.

Have we kept our promises and ignored the limitations of generative AI? Zvonimir says that, like any new technology, this one should be approached soberly and be well informed about what it is and what it is not. GenAI also has its weaknesses and limitations. Slowness. “GenAI is actually very slow. He can't follow the world in real time like a man. The only way you can communicate with him is through text. You need to write an instruction (prompt), then in a few seconds an answer arrives. After that, you write a new prompt (sub-question) and so on. GenAI will never be used, for example, in autonomous driving, as it cannot keep up with the complexity and dynamics of real-time traffic." The Apparent Authenticity. “When ChatGPT appeared at the end of 2022, people got scared because it behaved quite authentically. He seems to be talking to you as a man. Something like that might not have been expected. But it is still an algorithm, a tool through which we can perform some tasks better.” Computation – resource consumption. “To produce only one image of 1024*1024 pixels, the genAI consumes as much electricity as a single charge of an average smartphone. Imagine how much energy it would take to generate dozens of images per second, which is needed for e.g. video generation.”