It is no longer possible to ignore generative AI. It is better to sit around the table and try to understand how it could support our business and our work. But first, we need to be clear about how this disruptive technology works and what the opportunities and risks are. Such an overview will allow us to understand how each industry could improve thanks to generative AI, and how to make the most of the benefits of ChatGPT’s conversational chatbot for business.
What is generative AI and what can it do?
Artificial intelligence is a familiar field. We have addressed the topic several times in this blog, as it is a technology that can be applied in various contexts and that is offering considerable support to all professions. We have looked at AI’s ability to see, think, and act; its applications and its ability to analyze huge amounts of data; how to evaluate its potential use in the enterprise; and the experts to hire to ensure a successful implementation.
Now it is time to delve into an aspect that has been intriguing professionals, users, and even many students in recent months: generative AI. This is the ability of artificial intelligence to create content such as text, images, sound, and video using machine learning algorithms. Unlike traditional AI models that are trained to perform specific tasks (image recognition or text translation), generative AI seeks to create new content autonomously.
How is generative AI created, and how does it work?
- It starts with training data and its necessary cleaning and selection: generative AI learns to create new content by analyzing large amounts of existing data. For this, in addition to selection, the process of cleaning the raw data, which reduces quantity in favor of quality, is crucial. This process involves a series of steps, including the editing and deletion of incorrect and incomplete data fields, the detection and elimination of duplicate information and unrelated data, as well as the correction of formatting errors, missing values, and spelling mistakes.
- Then there is the neural network (Deep Learning): a mathematical structure that mimics the functioning of the human brain, allowing generative AI to learn and continuously improve over time.
- Finally, there is the machine learning model: this is the ‘brain’ of generative AI, which analyses training data and learns to generate new content autonomously.
All these elements are crucial to the functioning of generative AI, which enables machines to create content.
Such a huge opportunity could not escape the digital giants. Google Cloud is focusing its efforts on artificial intelligence, introducing tools such as Vertex AI and Generative AI App Builder. Software used to automate business processes and improve customer interactions. Microsoft, on the other hand, is extending OpenAI models to US federal agencies via Azure, paving the way for chatbot technology also in the government sector.
While generative AI can be a valuable asset to your business, it is important to always monitor its performance and ensure it is functioning properly. Click To Tweet
ChatGPT: opportunities of OpenAI’s generative AI model
Generative AI has grabbed the headlines and the interest of professionals and lay people alike ever since OpenAI made its generative intelligence model GPT (Generative Pre-trained Transformer) accessible. Based on language transformation (Transformer) and trained on a huge amount of text from the Internet, GPT is, to date, the most widely used generative AI model. The efforts of engineers and scientists in the training phase have given the algorithm extensive knowledge of the human language and the ability to generate coherent and convincing texts.
I asked ChatGPT what it can do, and its answer was:
” GPT is designed for human interaction through chat or text messages. It can answer questions, provide information, offer suggestions or even participate in informal conversations. ChatGPT uses the context provided in the conversation to generate relevant and coherent responses.”
After the praise, however, it also admitted its limitations:
“However, it is important to note that generative AI, including ChatGPT, has some limitations. Since generative AI does not possess a true understanding of the world like humans do, it may generate content that seems coherent but is actually inaccurate or misleading. Furthermore, generative AI may be susceptible to assuming biases in the training data, which could affect its responses.”
There are other conversational chatbot models that, in principle, offer the same output, but as ChatGPT is the most popular to date, we will use it as an example.
The risks of ChatGPT: from hallucinations to ethical aspects
Indeed, we must never forget that we are talking about a technology that has been trained by humans on such a large amount of text that there is a risk of “hallucinations”. Indeed, it is possible for the model to generate inconsistent, irrelevant, or even bizarre answers.
To overcome this problem, we must always fact-check the information. A thorough check of dates, names, and anything else we do not know. However, this is not the only risk of generative AI. Other equally important limitations concern not only the practical but also the ethical and social spheres. Here are a few examples:
- Law: from a legislative point of view, generative AI creates quite a few problems with regard to the violation of privacy and copyright. Suppose in the first case the Privacy Authorities are already regulating the aspects related to the violation of personal data. In that case, it is still difficult to understand how to act with regard to the attribution of work, since ChatGPT is not a natural person, and the results it generates are the result of a mix of elements from copyrighted works.
- Ethics: artificial intelligence lacks empathy. Empathy helps us develop an awareness of the consequences of our words and behavior on others. It prompts us to reflect on how our actions can intentionally or unintentionally hurt or harm people. In reality, not all humans are endowed with it in the same way, but artificial intelligence lacks it for now. Therefore, it can generate content that reflects stereotypes or prejudices related to culture, race, gender or other factors. It can provide offensive and inappropriate responses, although much work is being done on filters to mitigate this risk.
- Society: ChatGPT addiction and abuse, especially in schools, can increase backward illiteracy and atrophy our creative, numeracy, and writing skills. In addition, we will train our critical thinking less and less, and this will make us increasingly unable to distinguish false from true information. The entire school system will have to change to ensure that we learn from students who master AI tools, in some cases, better than teachers.
We cannot choose not to use generative intelligence, but we can do so consciously and responsibly. Its role will be to support certain activities, but without allowing us to lose control over the generated result. Furthermore, if we refer to the corporate sphere, it will be necessary to adapt and customize the technology to the culture and values of each productive reality.
How could ChatGPT improve activities in various business sectors?
We can have all the fun we want writing Shakespearian-style greeting cards, or composing poems from two simple words, or even making funny pictures like the Pope wearing a Moncler, but the really disruptive impact of this technology is in the professional sphere.
From writing emails or PowerPoint to improving customer service, ChatGPT is not only improving business but also creating new business opportunities. Let’s see what kind of support it offers to the main business sectors.
Marketing and sales: how to improve customer relationships
Obviously, the first area to benefit is content-driven marketing. With ChatGPT, you can get support in creating advertising messages, social media posts, and blog articles. The text generated can be turned into copy, but also into a script for videos or a prompt for images. The risk of using text indiscriminately, without the control and transformation of a professional, is the loss of corporate identity. Generative AI can work on style, but it is still far from emulating the corporate tone, which is made up of perceptions, images, and values associated with the brand.
It could, however, deliver satisfactory results in the creation of product user manuals consisting of lists of activities and actions, where a simple, detached tone is perfect for achieving the purpose.
The analysis of large amounts of data is definitely the strength of generative AI. With ChatGPT, it is possible to improve Customer Satisfaction thanks to its ability to analyze consumer feedback. It can also answer questions from potential customers and gather useful information for customer profiling. It could also analyze sales data and customer behavior to make predictions and suggestions for sales strategies.
Human resources: how to improve the selection and training phase
Hiring human resources is a delicate process. In fact, beyond the technical requirements and the soft and hard skills, there is the empathic side of the interview, the side made up of non-verbal communication, where the body emits signals that it does not govern. That is why, even in the field of human resources, we need to talk about support and not complete delegation.
ChatGPT could provide immediate answers to employees’ internal questions about company policies, benefits, recruitment or leave processes, and internal procedures, for example. Or take care of the very first stage of the training process for new employees by providing them with detailed information on company policies, organizational structure, internal resources and operating procedures. A sort of basic training to be then complemented by confrontation with the people who remain, to this day, the true essence of a company.
Generative AI can also be used as support in the first phase of the selection process. It could do an initial screening of CVs on the basis of pre-established macro-criteria, thus ensuring that only those candidates in line with the discovered position reach the physical interview with the human resources team.
Operations, IT and production: from anomaly analysis to customer support
Generative AI is able to offer support along all the stages necessary for the transformation of a project idea to product/service production. And in many cases, it also improves customer relations in the after-sales phase.
Starting from the research and development phases, the use of ChatGPT allows us to analyze large amounts of data and provide useful information, such as the identification of market trends or the production of technical reports. Not only that, it could also contribute to the creation of new ideas and concepts, providing inspiration and creative suggestions.
In manufacturing, artificial intelligence has long since made its way into making large and small devices intelligent. In addition to its benefits in data analysis, its ability to generate text has improved user interfaces. It has also provided insights into production planning, optimizing time and processes. Finally, it has made it possible to identify errors, anomalies, and production defects.
What support do you offer the IT department specifically? ChatGPT could provide support in solving common technical problems, such as software configuration or solving system errors. In addition, it could offer technical advice on the implementation of new IT solutions or the integration of existing systems. But it could also write code or auto-compile data tables using its ability to generate synthetic data.
Coming then to customer service, generative AI has made chatbots intelligent. Virtual assistants sped up communication with customers, especially in the information or sales assistance phase. Furthermore, ChatGPT could collect feedback from customers through surveys or virtual interviews, enabling companies to improve their products or services. As we can see, this generates a circle, as this information collected downstream will feed the research and development department.
Regulations, standards, and legal documents that need to be complied with: keeping up with the huge amount of legal information produced on a daily basis means spending hours reading long and complex legal updates.
Generative AI is capable of drafting and reviewing legal documents highlighting significant business changes that are worth reading. Not only that, if questioned on legal issues, it could answer questions about legislative aspects. These answers are linked to its ability to read and understand a large number of documents.
Internal communication and optimization of employee work
In general, algorithms such as ChatGPT are perfect for optimizing internal and external employee communication. From creating company presentations to summarising meetings, video conferences, or industry speeches.
Not only can ChatGPT act as a virtual assistant for internal procedures. For example, it can provide step-by-step instructions on how to submit a cost request or how to initiate an onboarding process for a new employee. This greatly simplifies the workflow. I will not need to wait for a colleague to be available to complete a task or to train me on standard company procedures.
As with all the activities we have discussed in this article, internal communication, in particular, requires good planning and proper management of the initial training phase of the model. This activity should be customized according to company guidelines. Generative AI support is very powerful but needs constant monitoring. As image and information damage can be just as influential.