Artificial Intelligence: Opportunities and Challenges

How AI is shaping our jobs and our society

When OpenAI launched GPT 3.5 in March 2023, I started experimenting with it. It took me a few interactions with this Large Language Model (LLM) to realize that it was a great improvement over existing assistants and bots.

Unlike Google Assistant, Alexa or Siri, ChatGPT was the first tool I interacted with that resembled human intelligence. It is capable of maintaining a conversation with follow up questions, understand nuances in the language and reply with highly coherent sentences. Its general purpose and user-friendly interface made it more popular than other artificial intelligence platforms like IBM Watson.

During 2023, we witnessed a dramatic increase in popularity of AI tools like Dall-E, Midjourney, Github Copilot and Bard. Some people changed their social media profile pictures for AI-generated avatars. OpenAI made it easy to create and customize your own GPT. Google’s Gemini and some devices like the hu.ma.ne pin or other weird gadgets will be released in 2024.

As AI tools become more sophisticated, they offer unique insights and solutions, transforming how industries operate and how professionals engage with their work. This integration of AI into the workplace is not just about efficiency — it’s about harnessing the power of advanced analytics and machine learning to foster innovation and creativity.

Where is the value of AI?

Technological change makes things cheap that were once expensive. Computer microprocessors made arithmetic cheap. The rise of the internet was a drop in the cost of distribution, communication, and search. The value of AI comes from a drastic reduction in the cost of predictions.

With artificial intelligence, we take the information that we have, to fill the information that we don’t have. It may not seem like much at first but, if we think about it, the ramifications and implications of this are vast. When we ask ChatGPT something, it takes the information that it has (context) to make a prediction:

What would a human do?

This low cost prediction capability can be extended to multiple areas, making it possible to democratize services that today are too expensive for most people, like access to healthcare or education.

How does AI impact software development?

Integrating LLMs in software development workflow in the form of coding assistants can potentially increase productivity of a development team, because it can accurately (in most cases) predict the intention of the developer and help fill in the blanks. Github Copilot and AWS Code whisperer are great examples of coding assistants that can be integrated in the development environment (IDE). Observability platforms like New Relic are increasingly integrating AI to help IT professionals get deep insights about the applications they are responsible for. These tools help teams make better decisions and become more productive.

Does this mean developers do not have to write code anymore?

No. Software engineers play a crucial role in the success of the projects they work on. It is the responsibility of the software engineer to write code that adheres to standards, is well written, secure and bug-free. Building great software depends on human expertise and critical thinking.

In computer science, we use abstractions. Assembly language is an abstraction so we don’t have to type 1s and 0s. High level programming languages like Python or C++ is then an abstraction on top of Assembly. Going up the layers of abstraction we will find Domain Specific Languages (DSL), Frameworks and code generators. If we keep going in this direction we lose flexibility and the ability to customize and tailor applications to specific business needs. Ultimately, it is the responsibility of the software engineers and architects to ensure that the building blocks integrate well together into a coherent system that satisfies business needs.

Will AI replace software developers?

AI can write functions in any programming language, find bugs, fix errors and write documentation. AI does not get tired or take sick days. But, as of now, AI does not possess intuition, creativity and common sense. In my opinion, AI will not replace developers anytime soon. Instead, there will be a shift in how developers spend their time. Repetitive tasks, boilerplate code and simple functions can be removed from our to-do lists. Problems that have been solved somewhere else, with well defined requirements are perfect for AI.

Real business problems are not that simple. Conditions are always changing, problems are ambiguous with undefined constraints, unique requirements, little available data and lots of uncertainty. This is the type of scenario where humans shine and that’s where software engineers will focus. The way I see it, there is an increasing collaboration between AI and humans, where AI takes care of the highly predictable tasks, and humans focus on innovation, creativity and exploring solutions to new unique problems.

What risks does AI pose to humanity?

Will there be robots in the streets destroying humanity with powerful laser guns? I doubt it. The real threat comes from the fact that Generative AI can replicate human language. LLMs can create coherent stories that may sound real but could be totally fake. AI gives the capability of deep faking people’s voices. Social media is already full of fake news and their algorithms decide what to show you. Humans are biased and, whether we accept it or not, not very rational. Our decisions are largely influenced by emotions. Most people don’t know what they want unless they see it in context. That context is defined by the news and information we are exposed to, as well as the people (and AI agents) around us. A reliable way of making people believe in falsehoods is frequent repetition, because familiarity is not easily distinguished from truth. If we combine all these facts about AI and our human nature, we could be immersed in a curtain of illusions created by AI. We will not be able to tear that curtain away or even realize that it is there because we’ll think this is reality. This could put our public conversations and our democracy in jeopardy, or even lead to wars.

What to do about the risks?

Should humans stop developing AI? I don’t think so, for two reasons:

1. Malicious actors will continue developing AI for bad purposes. Bad AI will only be stopped with good AI.
2. When put to good use, the benefits that AI brings to humanity are just too important to stop them. It can help fight disease, poverty, climate change and many other big problems that we are facing today.

Instead, what we need is regulation. We need to get involved and educate ourselves about the risks. It is important that governments, academia, businesses, and civil society work together to navigate these risks, which are complex and hard to predict, to mitigate the potential dangers and ensure AI benefits society.

Conclusion

Technology improvements around AI bring a lot of value, but they come with risks. It is important to educate ourselves and start taking action. It’s crucial to consider how we can best leverage AI to enhance our creativity and innovation.

References

• Agrawal, A., Gans, J., & Goldfarb, A. (2018). Prediction Machines: The Simple Economics of Artificial Intelligence. Harvard Business Review Press.
• Ariely, D. (2008). Predictably irrational: The hidden forces that shape our decisions. HarperCollins Publishers.
• Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
• AI is about to completely change how you use computers, by Bill Gates
• AI for software development: A reality check | Birgitta Boeckeler | LeadDev Berlin 2023
• AI and the future of humanity | Yuval Noah Harari at the Frontiers Forum
• Frontier AI: capabilities and risks
• What every CEO should know about generative AI