Imagine that you are a lawyer preparing the defense for a highly complex case you are working on. It's 2027, and instead of having to search through Google or your printed books for all the laws and articles relevant to your defense, you have a conversation with Cicero, your generative AI legal bot. Drawing on its specialized knowledge of laws and past cases that no human could remember, Cicero becomes your assistant to brainstorm the best defense strategy.
Cicero challenges you to think more deeply, helps you gather content, and, like a professional chess player, assists you in anticipating possible counterarguments from the prosecution. This thoroughly prepares you to defend the case much more effectively than the traditional method of consuming content and studying analogically.
The only significant challenge?
The prosecutor's attorney did the same and used Cicero as well. Cicero was developed by a startup that optimized OpenAI's language model to better suit the world of jurisprudence. It has accumulated numerous interactions, allowing the model to learn from each one while continuously improving and expanding its knowledge. Cicero has become a unicorn, and similar businesses are emerging in various sectors.
that is exactly the type of businesses that the world of AI will generate, and that we will address in this article.
We will discuss these new business models in the world of Artificial Intelligence and how we can innovate better by sector. This is precisely what Sam Altman, CEO and co-founder of OpenAI, the company behind Chat-GPT, discusses in the following statement. Watch the video below and remember to enable subtitles.
"I believe there will be a few big, fundamental models out there on top of which other people will build, but right now what's happening is a company creates a big language model, an API, and other people build on top. I think there will be an intermediary layer that becomes really important, where - and I'm a bit skeptical of all these startups trying to train their own models, I don't think that will continue - but what I think will happen is there will be an entirely new group of startups that will take a very large existing model from the future and tailor it (not just fine-tuning, but all the things you can do - a lot of access will be provided) to create the model for medicine, using a computer, or the friendly model, and so on, and these companies will create a lot of lasting value because they will have a special version that won't need to create the basic model, but will have created something they can use just for themselves or share with others, which has this unique data flow that improves over time and so on. So, I think a lot of value will be created in this intermediary layer."
In 2015, after completing their medical residencies, three young doctors from Rio de Janeiro decided to take a sabbatical year to focus on PEBMED, a technology company in the medical field. They joined a startup accelerator in Rio de Janeiro and decided to merge the best features of their apps into one. This led to the creation of Whitebook, an application that condenses all medical information and makes it incredibly easy to access.
The app gained popularity among medical students and even among doctors, turning the company into a huge success. In fact, in July 2021, Afya, a major educational group in medicine, acquired 100% of the company for R$ 132.9 million. It was a remarkable success, and I even had the chance to meet Bruno Lagoeiro, the CEO and founder, during my time as Director of Tinder in Rio. I truly admired his ability to excel not only as a physician but also as a businessman.
So, what was the key behind Whitebook's success?
It was the ability to leverage a technology that was rapidly growing at the time – mobile applications – in the field of medicine, and fundamentally improve the customer experience. If, at the time, the breakthrough was to develop apps for specific segments, now, as we also heard from Sam Altman's statement, the new breakthrough will involve sector-specific solutions empowered by Artificial Intelligence.
Allow me to explain this with an example from ChatGPT. In March 2023, OpenAI launched developer APIs for the ChatGPT model. The ChatGPT API utilizes the "gpt-3.5-turbo" AI model and allows developers to integrate versions of the popular chatbot into their applications, either unmodified or customized.
Snap's My AI, part of Snapchat's platform, is already one of the first examples of API usage. It's being used in conjunction with a new virtual tutoring feature for the Quizlet online study tool and a future Ask Instacart tool in the popular local shopping app. It's worth noting that the API is not limited to brand bots imitating ChatGPT but can be used to power "non-chat" software experiences that can benefit from AI.
OpenAI announced that the ChatGPT API would cost $0.002 per 1,000 tokens, which is roughly equivalent to 750 words. Additionally, the company is offering a dedicated capacity option for developers with higher budgets who expect to use more tokens than the limit provided by the standard API.
The possibility of building on top of APIs provided by companies developing Large Language Models like OpenAI creates extraordinary opportunities to revolutionize segments of the economy. Every leader, entrepreneur, or businessperson should take advantage of this.
Let's use the example of a chemical company or the chemical sector in general, which is still quite traditional. Similar to how Whitebook improved access to information, now, if you are a chemical company, you could have algorithms at your disposal that help you innovate more easily.
But, how?
Examples come from two impactful areas:
- Product Development:
With the ChatGPT API, you can generate new product ideas or enhance existing ones by analyzing customer feedback, identifying emerging trends, and exploring new applications for your products. For instance:
- Analyzing customer feedback: The API can be used to analyze customer reviews, social media posts, and other sources of feedback to identify common themes and pain points. This helps you understand what customers like and dislike about your products, generating ideas for new features or products that cater to their needs.
- Identifying emerging trends: The API can be used to analyze industry publications, social media posts, and other sources of information to identify emerging trends and technologies. This keeps you ahead of the curve, allowing you to develop products that meet the ever-evolving needs of customers.
- Exploring new applications: The API can be used to analyze scientific publications and other sources of information to identify potential new applications for your products. This helps you expand your product offerings and enter new markets.
- Research and Development:
With the ChatGPT API, you can generate new research ideas or explore new research areas by analyzing scientific literature, identifying knowledge gaps, and generating hypotheses for new experiments. For example:
- Analyzing scientific literature: The API can be used to analyze scientific publications and identify patterns and trends in research. This helps you identify areas where new research is needed and generate hypotheses for new experiments.
- Identifying knowledge gaps: The API can be used to analyze scientific publications and identify areas where knowledge is lacking or conflicting. This helps you focus your research efforts and prioritize areas that require further investigation.
- Generating hypotheses: The API can be used to generate new hypotheses for experiments based on available scientific literature. This assists you in developing new research ideas and exploring new research areas.
The truth is, this is already happening in other sectors. For instance, did you know that Insilico Medicine's AI scientists have been using generative AI since 2018 to design new chemical compounds that can be developed into new drugs to treat diseases? The generative AI chemistry platform by Insilico Medicine is one of the most exciting applications of AI in the healthcare sector.
Chemistry42 is an automated machine learning platform for drug design capable of finding new lead-like structures in just days. The platform connects state-of-the-art generative AI algorithms with medicinal and computational chemistry methodologies to generate new molecular structures with optimized properties.
Recently, scientists from Insilico Medicine published an article describing how they use generative AI to quickly design new molecular structures targeting proteins that play essential roles in disease progression. The article, titled "Chemistry42: An AI-Driven Platform for Molecular Design and Optimization," was published this week in the Journal of Chemical Information and Modeling.
Enormous opportunities exist to create new avenues, either within your company if you're a leader of a traditional enterprise, or by launching a new startup if you're an entrepreneur. To draw on the analogy of how refrigeration gave birth to Coca-Cola, the world of AI will generate the new Coca-Colas of the future that will emerge from the intersection of not only Artificial Intelligence but also its accessibility through APIs and traditional sectors of the economy.
