Navigating the AI Revolution: Lessons from Caltech

Back in May, I had the good fortune of attending a lecture on AI at Caltech with a friend. The lecture, entitled “Artificial Intelligence: The Good the Bad and the Ugly,” was presented by Yaser Abu-Mostafa, Professor of Electrical Engineering and Computer Science at the California Institute of Technology, who is also Chairman of Machine Learning Consultants LLC. This was, of course, a big deal to me, not just because I had been thinking about the implications of AI since ChatGPT first came into the public consciousness, but also because here was an opportunity to learn from the very best at one of the most prestigious universities.

The lecture included a technical but accessible description of AI, as well as an explanation on Neural Networks, the computational model underpinning Large Language Models (LLMs) like ChatGPT.

Because there is so much anxiety around the prospect of job loss and social impact, I am going to focus here on three key takeaways from the lecture that I think are relevant: 

1. There is no stopping the development of AI, but governments must regulate it; 

2. Notwithstanding the initial pain, there are tremendous benefits to AI applications, including better jobs in the future; 

3. There are enormous risks to the development of AI including malicious use by bad actors and adverse social impacts.

There is no going back, AI will reach its full potential. Currently, a number of AI experts, and very smart well-intentioned people, have voiced concerns over the speed at which AI is being developed and are therefore calling for a temporary halt. Prof. Abu-Mostafa explained that, however legitimate the concerns, this is not going to happen because of the “open source” community, which is made up of computer scientists and other experts who “openly exchange computer programs and ideas in order to build without restrictions” (Abu-Mostafa). At this point, then, any government can put together a team that will have access to this open-source information and can continue to develop on their own. If, say, the U.S. were to temporarily halt AI development, Prof Abu-Mostafa posits, all we’d be doing is giving the other guys, governments and bad actors included, an advantage. The thing to do, then, is to regulate AI by treating it like a product that requires government testing and approval before allowing it to reach the public. 

There is a lot to be gained in the development of AI, but fear that these new technologies will obliterate jobs are not unfounded. Compared to the Industrial Revolution which took approximately eighty years to complete its transition, AI is poised to take full hold in less than twenty years, according to Prof. Abu-Mostafa. The Industrial Revolution relieved manual labor with machines, and AI is relieving mental labor, making  jobs in all sorts of fields, including transportation, customer service, marketing, and even science, key targets. Now, historically, jobs that get taken over by machines open the door for new, better jobs. But before that happens, Prof. Abu-Mostafa expects we will see loss of jobs at a large scale, and offers a number of solutions, including a government safety net that will guarantee an income to those who get displaced, as well as training for current and future generations.

The deleterious implications of AI development are many but do not include rogue AI systems eradicating the human race. In essence, Prof. Abu-Mostafa proposes that instead of worrying about machines destroying humans, we need to address immediate dangers we know beyond doubt we’ll face, including cybercrime, disinformation, hacking, and scams that are similar to what we have experienced up to now but, as he puts it, “on steroids.” Beefing up law enforcement will be crucial, and making the use of AI in crime an aggravated offense will deter some bad actors.

Now, the social impact of AI is something that deeply worries Prof. Abu-Mostafa. He talked about the clear correlation between social media use and mental health problems, especially among the young. We are more isolated than ever, so the prospect that AI will replace friendships and human relationships in general is something to be taken seriously given our innate need for human interaction. 

If all this still feels a little abstract, here’s a way to think about it: Just as smartphones and social media have completely transformed our lives, AI will permeate all aspects of them, from what we see when we do a search on Google to the way we learn, work, shop, and interact with one another. For those whose jobs are threatened by AI, my advice is to learn about this stuff and explore how you can use it to your advantage. Lamenting the pitfalls of AI alone is not going to help us. As Prof. Abu-Mostafa humorously put it “If you can’t beat them, join them!” but do so in a way that preserves and nurtures your humanity. 

On that note, I want to disclose that, following in Prof. Abu-Mostafa's footsteps, I did not use ChatGPT or any other AI writing software to write this blog post. I did, however, use it to catch a few grammatical errors, and to pick the title of the post, which, I must confess, is better than what I could come up with.   

If you’d like to watch Prof. Abu-Mostafa’s lecture, and I highly recommend you do, here’s the YouTube Link.