From Promise to Peril: The Pros and Cons of Generative AI

By: Anirban Nandi – Head of AI Products & Analytics (Vice President) – Rakuten India

Most people associate ‘Generative AI’ with some type of end-of-the-world scenario. In actuality, generative AI exists to facilitate your work rather than to replace it. Its applications are showing up more frequently in daily life. There is probably a method to incorporate generative AI into your work, regardless of whether you operate as a marketer, programmer, designer, or business owner.

The potential and hype around artificial intelligence (AI) have become major topics. We can all agree that artificial intelligence (AI) has the potential to significantly improve our lives through tailored treatment, safer transportation, and a host of other applications. ChatGPT, while nifty, is merely the beginning; enterprise possibilities for generative AI are significantly more advanced.

According to McKinsey, generative AI, and other foundation models are revolutionizing the field of artificial intelligence, advancing assistive technologies, speeding up application development, and enabling advanced functionality for non-technical users. Most people associate ‘Generative AI’ with some type of end-of-the-world scenario. In actuality, generative AI exists to facilitate your work rather than to replace it. Its applications are showing up more frequently in daily life. There is probably a method to incorporate generative AI into your work, regardless of whether you operate as a marketer, programmer, designer, or business owner.

The article will focus on the pros and cons of Generative AI that is being spread across various industries.


One of the most significant impacts of generative AI on the text industry is its ability to automate content creation. With the help of natural language processing (NLP) algorithms and machine learning models, generative AI can create written content that is indistinguishable from content written by humans. This has the potential to reduce the cost and time required for content creation, particularly for businesses that produce a large volume of written content.


  • Large-scale language and image synthesis for automated content articles, blog posts, and social media postings can all be created automatically using AI models. For organizations and professionals that regularly develop the material, this can be a useful time-saving tool.
  • Variety of content is increased: Text, photos, and video can all be produced by AI models. Text, graphics, and video are just a few of the content forms that AI models are capable of producing. This can assist companies and professionals in producing more varied and captivating content that appeals to a wider audience.
  • Personalized content: AI models can create content that is tailored to the preferences of certain users. This can assist companies and professionals in producing content that their target audience is more likely to find interesting and, as a result, read or share.


We already know that these generative AI systems give rise to a number of moral and legal concerns very quickly. ‘Deepfakes’, or visuals produced by AI that appear to be realistic but aren’t, have already appeared in politics, media, and entertainment. A lot of issues over what counts as original and proprietary work are also brought up by generative AI. The makers of these systems contend that the generated text and images belong to their immediate creators because they differ slightly from any earlier content. However, it is obvious that they are derived from the earlier text and pictures that were used to train the models. All of these LLMs struggle with conversational skills. Owing to their exposure to previous human content during training, they have the propensity to repeat any racist, sexist, or otherwise prejudiced remarks.



In its most recent ‘Big Ideas 2023’ report, the investment management firm Ark Invest claimed that generative AI may result in a ten-fold boost in coding productivity. Moreover, generative AI can improve code reuse. One of the core principles of software development is to reuse code as much as possible to save unnecessary effort and speed up development. It is not always a simple process to reuse code because it requires locating the right section of code and customizing it to the needs of the current application.


While generative AI coding can generate code quickly and efficiently, it may lack the creativity and ingenuity of human developers, who can approach problems and solutions in unique and innovative ways. Its coding can produce code with errors and bugs that may be difficult to detect and resolve, requiring significant human intervention and expertise to fix. The coding can amplify and perpetuate existing biases and inequalities in the code it generates, potentially leading to discriminatory or unethical outcomes. It creates vulnerabilities in code that can be exploited by hackers and other malicious actors, creating potential security risks for software applications and systems.



The GPT-3 large language model created by OpenAI, which powers its ChatGPT chatbot and DALL-E picture generation services, along with the BERT large language model utilized by Google are examples of generative AI models. It can aid in data analysis and the comprehension of intricate systems. One of its major applications is the transformation of medical photos into photo-realistic visuals and the conversion of satellite images to map views to research new sites. OpenAI’s Dall-E is a free, albeit constrained, service that can produce unique and realistic artwork from text descriptions and natural language.


Image generators have been charged for violating privacy rights, copyright, and personal information. The text, images, and music generated by these tools are based on the earlier works of other authors and artists. There are restrictions with free services such as ChatGPT and Dall-E. The latter is likewise cost-free, however, each user is only permitted to create up to 50 free photographs in the first month and a further 15 images thereafter. Premium services provide better security and adaptability.


Although fascinating applications of generative AI have surfaced recently, primarily in speech-to-image creation using well-known models like Stable Diffusion and DALL-E, the technology’s commercial potential has largely gone untapped. And while both image and video have a place in business, speech is emerging as a strength.


Generative AI models can produce more natural and realistic speech than traditional text-to-speech systems. This can improve the quality of automated voice assistants, audiobooks, and other applications that rely on synthesized speech. It can be used to create speech for people who have difficulty communicating verbally, such as those with speech disorders or hearing impairments. This can help improve accessibility for these individuals and make it easier for them to communicate with others. For faster content generation it can make speech quickly and efficiently, making it useful for applications such as automated customer service, where speed and efficiency are important.


According to Mehrabian’s Rule, human speech may be divided into three components: words, tone of voice, and facial expression. Machine comprehension is text-based, and only recent advances in (NLP) have made it possible to train AI models on elements like sentiment, emotions, timbre, and other significant but not necessarily spoken components of language. While the analysis and AI synthesis processes can take some time, real-time speech-to-speech communication is often where it counts. Voice conversion must occur instantly when speaking is being done and translated correctly. Speech-to-speech technology must accommodate a wide range of accents, languages, and dialects and be accessible to everyone in order to realize its full potential. All users will need to support this AI infrastructure with thousands of different architectures for a particular solution because emerging technology solutions are not universally applicable. Additionally, users must plan for consistent model testing.


Machine learning algorithms called generative video models create fresh video data based on patterns and relationships discovered in training datasets. These models enable the creation of synthetic video data that closely resembles the original video data by learning the fundamental structure of the video data. There are numerous forms of generative video models, including GANs, VAEs, CGANs, and others. Each type adopts a different training strategy based on its particular infrastructure.


  • Efficiency: To create new videos fast and effectively in real-time, generative video models can be trained on enormous databases of videos and images. This enables the quick and inexpensive production of significant amounts of new video content.
  • Customization: Generative video models can create video content that is tailored to a number of requirements, including style, genre, and tone, with the appropriate modifications. This makes it possible to create video material more freely and adaptably.
  • Diversity: Generative video models may create a variety of video content, including films made from text descriptions as well as creative scenes and characters. New avenues are now available for the creation and distribution of video content.


Generative AI can produce unexpected results that may not be in line with the desired outcome. This lack of control can be frustrating and time-consuming to manage. Producing repetitive content or something that lacks diversity, as it can only generate content based on the data it has been trained on. The content produced can get very mainstream for the users. It can perpetuate biases present in the training data, resulting in biased video content. In the age of deep fakes, it can create videos that depict people or events that are not real, raising ethical concerns about the authenticity of the video content.


According to recent data, the global market for generative design technology is anticipated to increase at a compound yearly growth rate of 17.4% to reach $46.1 billion by 2025. Similarly to this, it is anticipated that the global market for creative AI will expand at a rate of 29.5% annually and reach $3.3 billion by 2025.


By automating numerous steps in the 3D modeling process, generative AI enables designers to produce more intricate and detailed models in less time. As a result, designers can produce more realistic and intricate 3D models, giving users more immersive experiences. can assist designers in exploring fresh design ideas and developing modifications of current models, resulting in more imaginative and cutting-edge designs. Generative AI can lower the cost of creating high-quality 3D models by automating several processes involved in 3D modeling.


The high computational resource requirements of generative AI approaches make them unfit for various applications. Models can occasionally create unexpected or challenging results, giving designers little control over the output and forcing them to manually alter or refine it. Even though generative AI models often claim to be accurate, this is not always the case, especially when working with large or highly detailed models. Some designers may find it challenging to embrace this strategy because it requires some level of competence in both domains to use generative AI in 3D modeling.

AI models have the potential to change a number of industries with the development of cutting-edge methods like GANs and variational autoencoders. It is crucial to remember that AI-generated content must be handled responsibly and with care for ethical and legal issues. Future generative AI applications are likely to be even more fascinating and creative as the technology develops.

Abraham Lincoln once said, “The best way to predict the future is to create it”. Generative AI has done just that and it’s only the beginning. It is booming and we should not be shocked. Many technologists view AI as the next frontier, thus it is important to follow its development. The potential applications of AI are limitless, and in the years to come, we might witness the emergence of brand-new industries.