Generative AI: To Infinity and Beyond
Welcome to the world where artistic avatars on social media created using the LensaAI app resemble works by Andy Warhol, classical music in the style of Beethoven can be composed by AIVA in less than a minute, and content written by ChatGPT3 exhibits quality more than sufficient to produce a book in just a weekend[1], appear in scientific journals[2], or excel in an MBA exam at Wharton School.[3]
For those working at Quidgest, who have been harnessing the potential of artificial intelligence for over 30 years to revolutionize the way digital solutions are created and delivered, none of this is particularly unexpected, surprising, or apocalyptic. It only seems so to those who haven’t yet prepared (considering how to make the most of its benefits) for this new era of Generative AI – which, in some cases, might not yet be perfect, but is steadily progressing towards it.
So, how does Generative AI work?
A generative artificial intelligence system is designed to create something new based on something that already exists. In other words, through machine learning, deep learning, neural networks, large amounts of data (often measured in petabytes, where 1 petabyte equals 1000 terabytes), and existing patterns within that data, AI can be taught and trained to perform specific tasks and generate new audio, image, text, simulation, video content, etc.
Taking the three examples mentioned in the introduction of this article: LensaAI was fed with thousands of specific design and artistic style information and rules that enable it to autonomously generate original images, not just copies of what it ‘knows’; AIVA (Artificial Intelligent Virtual Artist) was trained with thousands of examples of classical scores and music from different genres, rhythms, and harmonies, capable of composing new melodies based on the analysis of those samples; and ChatGPT (Generative Pretrained Transformer) was trained with thousands of examples and text patterns available on the internet, to produce a natural language model capable of recreating dialogues, letters, poems, emails, articles, and other types of narratives commonly produced by humans. The model was trained to predict words – the neural network was fed with text sequences and instructed to predict the next word in the sequence based on the previous words. This allows the model to recognize and reproduce complex linguistic patterns found in natural text.
Unlike more traditional approaches that rely solely on a large amount of data to be effective, Generative AI models can generate high-quality new data from a smaller dataset – because these models can identify underlying patterns and relationships in the data to generate new content.
The goal of this technology is precisely to replicate human behavior, creating based on what humans have created before. That’s why the results are so impressive: because the larger and better the data sample, the more patterns can be detected and the more variations can be created that align with what a human could achieve.
Creating, Innovating, and Scaling Businesses with IA
The widespread use of AI goes far beyond beautifying social media or speeding up text writing. This phenomenon applied to organizations promises to continue radically changing how they operate, innovate, and scale their business. In the book “All In on AI: How Smart Companies Win Big With Artificial Intelligence,” Thomas Davenport (Professor at Babson College) and Nitin Mittal (Head of US Artificial Intelligence Growth at Deloitte) consider three main archetypes for organizations’ use of AI: 1) creating new businesses, products, or services; 2) transforming operations; and 3) influencing customer behavior. The authors provide specific examples of global organizations that already use AI within these three archetypes to increase their productivity, profitability, and competitive advantage:
- Amazon: Amazon uses AI to personalize product recommendations for each customer and improve logistics processes by forecasting demand and optimizing delivery routes.
- Google: Through its search platform and voice assistants like Google Assistant, the company has been improving its ability to understand the context of user questions to provide increasingly precise answers.
- Netflix: The company collects data on its subscribers’ viewing habits and uses machine learning algorithms to understand their tastes and preferences, enabling highly personalized content recommendations (similarly done by Spotify with music playlists).
- Boston Dynamics: This robotics company develops robots capable of performing complex, strenuous, or hazardous tasks for humans, such as moving heavy objects in factories and warehouses.
- Tesla: Through sensors, Tesla collects information to train its algorithms and make its vehicles more accurate in detecting obstacles and making driving decisions.
Read more: In the era of artificial intelligence, let’s invest in human intelligence*
90% of Online Content Created by AI by 2025
Suddenly, writing movie scripts or drafting lease contracts has become much easier, and identifying the true author of artistic, academic, or journalistic works much more challenging. But this story is still in its early chapters. According to the latest Europol Innovation Lab observatory, [4]by 2025, it is expected that 90% of the content available on the internet will be produced with the help of artificial intelligence.
A McKinsey study[5] shows that AI adoption has more than doubled in the past 5 years. The Financial Times also reported that investments in Generative AI exceeded $2 billion just last year, highlighting the clear and growing interest from investors and companies in this technology[6]. Greg Brockman, one of the co-founders of OpenAI, also shared a tweet announcing the tremendous impact of AI not only on the economy but specifically on the job market: “This technology can already do all sorts of things that companies currently need humans to do. Even if this tech doesn’t take over your entire job, it might very well change it.”
Kevin Scott, Chief Technology Officer at Microsoft, also believes that 2023 will be the most exciting year the AI community has seen[7]. Here are some of the key reasons he makes this assertion:
- Limitless Creativity: As Generative AI becomes more popular and accessible, more people can use the technology to express themselves creatively in music, arts, writing, etc. These new tools also democratize access to new skills previously dominated by specialized professionals or highly experienced individuals in fields such as law, medicine, or history, among many others.
- Assistance at Work: How often have we needed support in repetitive, tiresome, time-consuming, complex, or ancillary tasks, in order to focus on truly strategic and valuable activities in our role? AI is now taking on this mission, allowing humans to gradually move away from all types of work that do not dignify or fulfill them.
- High-Speed Innovation: Generative AI significantly reduces the time for creative processes between the idea, the need, and the solution, the final result. This virtuous collaboration cycle results in a quantum leap in terms of innovation across all areas that is hard to achieve through traditional methods.
To Err Is Human, and So Is the Machine
This extremely powerful AI method is, of course, subject to failures. This is because AI models are built and trained based on data and algorithms originally developed by humans – and when these contain errors or are incomplete, the results generated by the machines might also be flawed.
Furthermore, AI models also do not (yet) have the ability to understand the context or nuances of human language, for example. Therefore, inappropriate, incorrect, or even offensive responses may arise – hence the need for human intervention to validate the outputs, according to parameters of current legislation, accuracy, truthfulness, suitability, ethics, and others.
The Importance of Asking Good Questions and Validating That We Obtain Good Answers:
- Bias: Generative AI models can replicate biases[8] present in the data they were trained on, resulting in discriminatory or offensive content. Disturbing or surreal images, such as hybrid animals or unclothed female bodies, and homophobic or racist texts are some examples.
- Accuracy: Due to a lack of factual knowledge, outdated data, or errors in understanding context, these applications can be used to produce fake news, deepfake videos, and other false or inaccurate content.
- Context: Generative AI models can generate irrelevant or inappropriate responses, especially when dealing with complex questions – inaccurate or humorous translations and voice instructions occur due to errors in context comprehension, lack of understanding of natural language, or lack of cultural knowledge.
- Imagination: While these models can produce astonishing original content, they are still limited by the quantity and quality of the data they were trained on. Hence, many artists do not appreciate poems, lyrics, music, or artworks generated by these tools.
- Regulation: Identifying and defining the originality or intellectual property of artificially generated content[9] and controlling its application and the risk of misinformation are topics that have been fiercely debated in the realm of AI.
- Security: The widespread use of Generative AI poses a series of dangers associated with hacking, such as fraud and identity theft, leakage of confidential information, manipulation of data in economic or governmental contexts.
Genio Platform: From Generative to Ingenious
Etymologically, the Latin root “gen-” refers to ‘giving birth’ and is present in various words in our lexicon such as generation, gene, or genesis. Generative AI draws on this Latin heritage, referring to a characteristic of something capable of generating or creating something new. In other contexts, the term may also refer to something with the ability to regenerate or renew itself constantly.
Considering the increasing complexity of software, the layers supporting an information system, and the number of technologies, components, and requirements needed for its development and operation, it’s safe to say that being a true genius is necessary to develop software today.
Genius is another word that, etymologically, refers to creation, and it’s the word that inspired Quidgest to create its own automatic software generation platform in 1990 – Genio. Just as we understand a genius as someone with exceptional or superior natural abilities or talent to create extraordinary things, Quidgest’s Genio is put to the test daily to meet needs, solve puzzles, and overcome challenges that many would consider impossible – transforming non-IT professionals into creators of digital solutions, updating obsolete systems in record time and quality, creating error-free software ready to evolve right after installation, protecting personal data and institutions against crimes and frauds with superior efficiency, supporting organizations with strategic management tools that allow them to achieve unprecedented levels of efficiency, quality, and innovation.
These are just a few of the characteristics of solutions designed through the Genio platform, in an agile Model Driven Engineering (MDE) environment which, using artificial intelligence, models, and patterns capable of representing complex knowledge, can be easily tested (through simulations, anticipating behaviors or predicting evolutions), enabling extraordinary increases in speed, efficiency, productivity, and reliability.
If we use the metaphor of Aladdin and the Genie in the Lamp, we can also refer to Quidgest’s Genio as the solution ready to be ‘unleashed’ by any professional or organization to meet their specific needs and challenges (and here more than three ‘wishes’ are contemplated!).
Here are 10 aspects that distinguish Genio from other platforms and tools of Generative AI:
- Technological Independence: Genio combines vendor and technology independence, standardized code, and fast delivery across multiple languages.
- Speed and Productivity: Genio automates around 98% of code and develops projects up to 10 times faster with 1/10 of the usual resources compared to traditional methods. It’s also 8 times more productive than traditional low-code platforms.
- Easy Evolution: Genio-built systems are easily adaptable, allowing companies to integrate new requirements and respond to constant changes quickly and efficiently.
- Customization: Genio creates specialized information systems tailored to the specific needs of demanding and diverse business areas.
- Integration and Interoperability: Genio is designed for seamless integration with other tools and systems, ensuring interoperability across different manufacturers and brands.
- Maximum Precision: Genio automatically generates high-level, error-free standard code that is independent of the generation framework, ensuring consistency and quality. Testing is integrated throughout the development process.
- Continuous Updates: The platform is regularly updated with the latest technologies, features, and work methodologies to provide accurate and reliable results. It also keeps the generated information systems up to date for sustained productivity.
- High Abstraction Level: The high-level graphical interface allows for building and maintaining information systems without the need for extensive programming knowledge. Genio shifts the focus from code to business patterns and models.
- Support and Training: Quidgest offers customer support and training to ensure maximum utilization of the capabilities of Genio’s solutions and features.
- Results-Oriented: Genio was developed to help individuals and companies achieve tangible results in terms of efficiency, productivity, agility, and cost reduction.
Comparison with GPT- 3:
Genio takes a more reliable approach to automated code generation compared to tools like GPT-3, which lack internal models of business processes, application data, or application logic. GPT-3’s unpredictability stems from its inability to construct these internal models, making it less precise and prone to errors.
Genio’s reliance on a combination of modeling and AI offers a more trustworthy solution. The modeling process captures system requirements and creates models that can be verified and validated before code generation. This approach results in more reliable and robust software systems.
Additionally, Genio boasts impressive speed, generating code at a rate of 1 million characters per second, enabling the conversion of complex models into readable code in just 2 minutes.
From AGI Renaissance to Transhumanism:
This New AI Spring[10] is blossoming visibly, and its impact spans across all fields and sectors. In the case of Generative AI, its application, in the very near future, will become a part of everyone’s daily life, not just a few, through mass content creation for the Metaverse and NFT space, fully automated production of movies, music, and books; the advancement of autonomous vehicles equipped with much faster and precise ‘reactions and abilities’ than humans, the evolution in creating new materials, ingredients, or medicines for novel diseases based on molecular modeling, compound design, and data analysis.
In line with the excitement that has always haunted Artificial Intelligence, we are rapidly moving towards AGI (Artificial General Intelligence)[11] and envisioning a future where artificial intelligence will serve to transform society and address some of humanity’s most critical issues, such as climate change, health and well-being, or social and political concerns. Some believe that AGI could be the next evolutionary step in the history of technology, and that it will revolutionize how people live, work, and interact with each other.
“By 2029, computers will have emotional intelligence and be convincing as people.”
Ray Kurzweil
This possibility of coexisting in all aspects of our lives with highly evolved and versatile machines (thanks to machine learning, computer vision, natural language processing, simulation of the human brain, etc.) will make computers capable of performing any human intellectual task – from telling a joke to expressing emotions, much like in Spike Jonze’s movie “Her.”
Ray Kurzweil, an inventor, futurist, and well-known Director of Engineering at Google, believes that the processing power and emotional intelligence of computers will be comparable to human intelligence around 2029[12]. At that point, we won’t merely ‘use’ computers. We will interact, converse, confide, and even form relationships with them. (Interestingly, Kurzweil’s predictions have had an accuracy rate of 86% to date).
We are also living in a dream era for Transhumanists, a movement inspired by the English biologist Julian Huxley in 1957. They advocate that human evolution shouldn’t be confined by the limits of our species. It should be enhanced through technological and scientific advances, making the seemingly impossible achievable. It’s worth noting that what was considered impossible in 1957, like GPS, the Internet, or mobile phones, is now part of our daily lives.
This isn’t just science fiction; it’s reality. From cyborgization (integrating technology into the human body to extend life, mobility, or strength, as with a heart bypass or a prosthesis) to the implantation of chips like those from Neurolink[13] to restore lost functions such as vision or enable information interaction with other electronic devices, and even uploading and downloading information into the human brain to create superintelligent humans or transfer human consciousness to a computational system, enabling people to ‘live forever’ in digital form.
Generative AI, combined with other disciplines like Neurotechnology or Genetics, has proven to be a key in creating innovative solutions for complex problems, allowing humans to achieve new levels of knowledge, creativity, competence, and longevity.
Quidgest aims to be a part of this journey in artificial intelligence through software. We aspire not just to continue, but to do more and better for individuals, businesses, public institutions, and the planet. Through information management systems that revolutionize disease prevention, new treatment development, professional training and development, access to affordable housing solutions, predicting hazards and financial trends, making informed decisions in justice and public policy, monitoring natural resources, and contributing to climate action… and much more across various societal sectors.
We want to go “to infinity… and beyond,” borrowing the iconic phrase from Buzz Lightyear in the movie “Toy Story.” And we want to do it driven not only by our ambition but also by the mission we’ve pursued with dreams, determination, and enthusiasm for 34 years: to dare, to be, and to make a difference every day.
[1] He Used AI to Publish a Children’s Book in a Weekend. Artists Are Not Happy About It. Time, 14 dezembro 2022: https://time.com/6240569/ai-childrens-book-alice-and-sparkle-artists-unhappy/
[2] Fake Scientific Abstracts Written By ChatGPT Fooled Scientists, Study Finds. Forbes, 10 janeiro 2023: https://www.forbes.com/sites/brianbushard/2023/01/10/fake-scientific-abstracts-written-by-chatgpt-fooled-scientists-study-finds/?sh=68b7f48d18b6
[3] ChatGPT passes MBA exam given by a Wharton professor. NBC, 23 janeiro 2023: https://www.nbcnews.com/tech/tech-news/chatgpt-passes-mba-exam-wharton-professor-rcna67036
[4] Europol Innovation Lab (2021). Facing Reality: Law Enforcement and the Challenge of Deepfakes. https://www.europol.europa.eu/cms/sites/default/files/documents/Europol_Innovation_Lab_Facing_Reality_Law_Enforcement_And_The_Challenge_Of_Deepfakes.pdf
[5] McKinsey & Company. (2022). The State of AI in 2022: And a Half-Decade in Review. QuantumBlack. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2022-and-a-half-decade-in-review
[6] Investors seek to profit from groundbreaking ‘generative AI’ start-ups. Finantial Times: https://www.ft.com/content/9c5f7154-5222-4be3-a6a9-f23879fd0d6a
[7] A conversation with Kevin Scott: What’s next in AI, Microsoft: https://news.microsoft.com/source/features/ai/a-conversation-with-kevin-scott-whats-next-in-ai/
[8] ChatGPT could be used for good, but like many other AI models, it’s rife with racist and discriminatory bias. Insider, 16 janeiro 2023: https://www.insider.com/chatgpt-is-like-many-other-ai-models-rife-with-bias-2023-1
[9] He used AI to win a fine-arts competition. Was it cheating? The Washington Post, 2 setembro 2022: https://www.washingtonpost.com/technology/2022/09/02/midjourney-artificial-intelligence-state-fair-colorado/
[10] The New Spring of Artificial Intelligence and Models. QuidNews 27, julho 2019, pp. 14-23: https://quidgest.com/wp-content/uploads/2019/09/Quidnews_27.pdf
[11] Artificial General Intelligence (AGI): What You Need to Expect. Future Side, 15 agosto 2022: https://futurside.com/artificial-general-intelligence-agi-what-you-need-to-expect/
[12] Computers will be like humans by 2029: Google’s Ray Kurzweil. CNBC, 11 junho 2014: https://www.cnbc.com/2014/06/11/computers-will-be-like-humans-by-2029-googles-ray-kurzweil.html
[13] Musk: Neuralink Will Reach Human Testing by Mid-2023. Business Insider, 27 julho 2022: https://www.businessinsider.com/neuralink-elon-musk-microchips-brains-ai-2021-2#neuralink-is-developing-two-bits-of-equipment-the-first-is-a-chip-that-would-be-implanted-in-a-persons-skull-with-electrodes-fanning-out-into-their-brain-2
