We have all read science fiction books and movies in which artificial intelligence surpasses human intelligence or succeeds in subduing man, but in reality, digital assistants based on conversational artificial intelligence can be beneficial for the further development of competitive societies. Some creative professions could be threatened by ChatGPT in the future, but even three decades ago there were certain professions that do not exist today as a result of technological development, that’s why new ones are created every day. These were just some of the conclusions that were developed at tonight’s youth debate “Is there an understanding with chatbots?”, held in “Public Room” in Skopje.

The question “Is there an understanding with chatbots?” was first addressed by Tamara Dimitrova, researcher and co-founder of the Educational Center Scamper.

“When we speak of robots and chatbots, we get the perception that they are very close to our capacities and capabilities as humans,” said Tamara Dimitrova.

However, Tamara clarified that even though science fiction books and movies portray this, in essence, we are still far from surpassing humans in artificial intelligence. She mentioned that AI (Artificial Intelligence) is a very broad field and is the broadest in computer science, that is, artificial intelligence is quite a broad field, and chatbots are only a small part that is gaining a lot of momentum.

The moment an artificial intelligence becomes conscious, and this is also a great philosophical moment and a matter of philosophical debate, it will create an artificial intelligence that will be more powerful than itself until a point of singularity is reached. If that is reached, then AI will be more powerful than human intelligence and then the question arises of whether AI can surpass us.

Where will the limit be? Tamara Dimitrova adds that from today’s perspective, we can talk about two periods – before and after the appearance of OpenAI. She makes a distinction between chatbots and Chat GPT, because in the period after the appearance of OpenAI we can get answers to complex questions, ChatGPT can write essays and even analyze text. Previously, chatbots were used to answer frequently asked questions, provide basic information about various companies on websites, or provide weather forecasts.

However, Tamara Dimitrova pointed out that even after the emergence of OpenAI, the digital assistant based on conversational artificial intelligence cannot answer questions with empathy and emotional intelligence, provide information and advice in the direction of finance and law, independently make medical diagnoses or understand sarcasm, irony, and humor from context.

Several ethical dilemmas arise from a such technological explosion, says Tamara Dimitrova. First, on the Internet, different people provide different sensitive information which can be used in an inappropriate way. The second problem is privacy, but also bias. Accountability and transparency are two other issues, along with the problem of chatbot addiction, which if used too much to get interactions and knowledge, which can lead people to be even more closed off. Another important issue is Internet security.

The story of chatbots begins back in 1966

As a follow-up, Nino Karas – CEO and co-founder of CodeWell Group and CodeWell AI – joined the debate. He pointed out that they have been working in the field of conversational artificial intelligence as a company since 2015-2016, while the CodeWell AI platform has been created in 2018.

In his speech, Karas looked back on the history of chatbots, pointing out that their story began back in 1966 at MIT University in Boston (USA), where a professor created a chatbot that aimed to play the role of a psychotherapist. As a simplified form of keyword recognition, he asked the interlocutors some questions and even then managed to deceive many people who thought that there was a real psychotherapist on the other side.

“The development of computer systems led to the development of chatbots in 2014-2015, which are expert systems that follow some predefined conversation scenarios that someone has previously determined. They are much simpler systems than ChatGPT. These are expert systems that follow a conversation tree and predefined responses,” says Nino Karas.

Over time, much more advanced conversational artificial intelligence (Conversational AI) was created. This artificial intelligence represents an interface that is actually the next step in the digitization of institutions. This step begins with personal computers, then in the 80s of the last century, the IVR systems are developed, changing the contact centers of many companies, then, in 1999 comes the introduction of web applications, and in 2012 mobile applications, so that we could reach conversational artificial intelligence today.

“There is an important distinction between chatbots and conversational AI,” says Nino Karas. As he adds, chatbots are an older generation of assistants that operate on pre-defined rules. You can draw some scenarios yourself or enter keywords or phrases that the chatbot needs to recognize, but in the end, you get limited answers and they often answer: “sorry, I don’t understand you” or “sorry, I’m a robot”.

Among the chatbots, Karas singled out the example of DoNotPay, which, as a kind of lawyer, has helped a huge number of users in the US to get refunds for parking fines and managed to refund several million dollars of fines.

Unlike chatbots, Karas emphasized that ChatGPT is different because it is an advanced digital assistant with open domain knowledge, that is, you can ask them anything, unlike those with closed domain knowledge, who have limited knowledge only about products and services of a particular company.

“He can explain very advanced topics, solve math problems, he can write code, write music in any genre, write blogs, essays and you can discuss any topic with him. Chat GPT reached 1 million users in just five days and its potential is huge,” says Nino Karas.

So far, ChatGPT has managed to pass an MBA exam at the Wharton School of the University of Pennsylvania, the biggest CEOs use it to return emails on their behalf, but also many people have started to be afraid that they will lose their jobs, such as copywriters, people who work in journalism or write creative content.

Karas shared with the audience the experience of creating Slavcho from A1 Macedonia, as the first digital assistant based on artificial intelligence that understands and speaks the Macedonian language. From the sale of phones and services, to providing information to users similar to those you will receive from the contact center, Slavcho today serves about 40,000 users on a monthly basis.

“There are not many chatbots in our country and I am honored that we are starting to work in the public sector as well. A digital assistant ADA will be launched soon, which will automate communication in four state institutions”, said Nino Karas.

Photo: Metamorphosis Foundation

In the EU, artificial intelligence must not violate privacy

Marijana Jancheska – specialist for openness and participation at the Metamorphosis Foundation, joined the debate with her own speech. She pointed out that there are binding and non-binding documents that regulate the issue of artificial intelligence, and what is important right now and is in the process of being adopted is the first European regulation on artificial intelligence, which will be binding on all artificial intelligences used by citizens of EU member states.

Jancheska mentioned that Article 22 of GDPR (General Data Protection Regulation) is also important. In terms of chatbots, she mentioned that we can talk to them, but the way the legal entity has the artificial intelligence, the way it manages the information that we will share, must not violate our privacy, that is – we should be protected by them.

Marijana Jancheska added that within the European regulation, artificial intelligence is divided into several classes, namely: AIs that have a high risk of being misused, AIs that are known in advance not to be used, and those AIs with low risk.

She cited several examples where artificial intelligence cannot be used. For example, banks may not use it when processing customers who have applied for a bank loan.

“The police must not use AI to profile you as a potential perpetrator of some crime in the future,” Jancheska points out.

On the other hand, artificial intelligence can be used in the hiring process and in the process of providing medical diagnoses.

Marijana Jancheska added that the Metamorphosis Foundation is currently part of the working group for the Artificial Intelligence Strategy, which is being developed. In the country, we do not have institutions for artificial intelligence production, although there are such institutes in Albania and Serbia.

In her speech, she pointed out that the European regulation on artificial intelligence has been in the works for several years, it has to go through several committees and the Council of Ministers, and later there will be an adaptation period of two years.

“However, two years is a very long time for technological innovation. From November until now, a million practical uses of ChatGPT have taken place. Technology will develop to a certain level at which the law will not be able to follow it”, says Marijana Jancheska, adding that the technological possibilities are so great that the legislator cannot exhaust all the scenarios for everything that technology can bring.

However, in response to one of the audience’s questions, Tamara Dimitrova added that in some countries, digital assistants based on conversational artificial intelligence in the field have been removed from the list for use for the time being, that is, ChatGPT cannot be used because there are too many different legal regulations in different countries.

Nino Karas added that ChatGPT is not connected to the internet and cannot know real-time data. For example, it cannot know what the current price of Coca-Cola stock is, and it cannot advise whether to buy the stock. And the question of economy is who will be to blame if ChatGPT gives the wrong advice.

“It will not substitute people any time soon. We will have changes in many professions. It will change creative professions, so I wouldn’t be surprised to see AI music composers”, answered Nino Karas to one of the questions from the young people in the audience.

Tonight’s debate “Is there an understanding with chatbots?” was organized by the Metamorphosis Foundation in collaboration with the framework of the project “Increasing Civic Engagement in the Digital Agenda – ICEDA”, co-financed by the European Union, and the project “Reliable News for Empowered Youth” financed by the British Embassy in Skopje.

Link to the original text: ChatGPT shouldn’t be seen as a danger, but as a new development opportunity |