Gilad Barkan who leads Wix’s Data Science guild gives us a glimpse to an AI driven future.

Hi Gilad. It’s been a while since the AI revolution swept the world - were you overwhelmed as well?

Actually, not at all. AI has already been around for quite some time, including at Wix. Wix's first AI based product was launched back in 2016. Since then, Wix has built a top class AI organization, consisting of dozens of data scientists and AI researchers. The AI infrastructure we've built made it possible to deploy 200+ AI models to production in over 20 verticals.

How did AI change the life of Wix users?

In 2016 we already saw AI as an enabler that can do much of the website building heavy lifting for our users. Back then it was very premature, but since then the technology and abilities of AI have changed dramatically. We deeply understood the potential of AI and saw a great response from our users, who loved this experience and adopted it exponentialy. It wasn’t even ChatGPT-based, because back then we didn’t have ChatGPT. We immediately understood the tools this technology hands us, and we’ve made great progress since then.

Nowadays 2016 seems like prehistory. Can you walk us through what is expected in the future?

We’re still at an early phase of this Big Bang. Trying to envision what AI will look like several years ahead is extremely challenging. Once we reach the point of AGI (Artificial General Intelligence) with true human-like reasoning capabilities, everything will change - not only our industry, but all of society and humanity is going to be completely disrupted.

How commonly will it be used?

It will probably be in the same capacity as cell phones. Today there are more cell phones on Earth than humans - the same will be true with AI models. It will be more and more specific and tailored for each user, and eventually there will be more AI models than human beings. It will be a trivial part of our life, either as a personal assistant or an enabler for industrial use for all walks of life.

So this knowledge won’t be kept only in the hands of technical wizards?

Just like software moved towards open source, I predict we’ll see a movement that broadens towards open source in a way that will reach the broad public. Not only tech giants will have control of it - it will have to be accessible to the public through open source, commoditized and specialized smart experts.

There’s obviously this common fear that AI will replace humans.

That’s the social challenge. Everyone is afraid of losing their jobs to AI and at some capacity, in specific tasks, some of us might - but in general, AI will serve more as our assistants than our replacements.

It's an understandable fear.

With AI, I believe we can achieve significantly more with the same workforce, enhancing efficiency without causing mass redundancies. It's about pushing the boundaries of our capabilities in a world that constantly seeks growth.

From what I gather, the world of Generative AI is moving towards micro-specialization. If I currently need assistance in a very specific trade or business, ChatGPT isn’t very helpful. Excuse me for sounding anxious, but shouldn’t this thing be figured out already?

It should and it already exists to some extent, but there are still these things called hallucinations (misleading results generated by AI). It’s caused by a variety of reasons. You need to remember that AI is based on an average of all the knowledge it contains and it’s trained to predict the next most probable word out of this knowledge - it doesn’t know whether the information is reliable, serious and trustworthy. Some of the information may have been posted by anonymous online commentors. It isn’t bullet-proof fact-checked information. That’s what causes these hallucinations. Sometimes there just isn’t enough information, and LLM (Large Language Model) isn’t programmed to say “I don’t know” so it spits out the best it can. So obviously there’s an inherent problem of trustful knowledge. If you want it to be 100% trustful, there’s still some work to be done. For example, we know that governmental organizations are already using ChatGPT to produce position papers.

Why is that alarming?

Because it’s not trustworthy. At the end of the day, these machines are quite dumb. We think they’re smart, but they’re just as smart as they’ve been fed - just like humans. They can give me many good reliable answers, but if one of every ten answers are hallucinations - well, it’s dangerous and it can’t be trusted. Gen AI isn’t consistent. I can ask the same question twice and receive two completely different answers, so which one of them is the truth? One of humanity's most distressing qualms in this era is “What is the truth?” Truth becomes more fragile and volatile. The challenge is to make GenAI more trustworthy. If it won’t be trustworthy, it will assist humans only partially.

So, what’s the next big step for Gen AI?

We want to make AI super-human. Real intelligence is beyond knowledge and that’s the promise of Artificial Intelligence. Now, how do I predict something I have never seen before? You need real intelligence for that. Predicting something that is out of the distribution - now that’s a real challenge.

How do you envision AI transforming the work we do in Wix in the coming 5-10 years?

Oh, there are plenty of ways in which GenAI could uplift our business. For example, moving from DIY to a Do It For You (DIFY) AI assistant, using emerging technologies like agents and tools to make this possible. Another great upside of this revolution has to do with website creation and design: as a business and website platform, our product involves multi modalities like text, images, design and layouts. We definitely see the GenAI technologies progressively merging these modalities together more natively.

We obviously know that GenAI is shaking the ground and truly gives new functionalities to basically everyone. However, if companies didn't truly invest in AI before this dramatic point in time, they'll soon face the fact that it will take them some time to execute a true change using AI within their products. We’re already reaping the fruits of the AI seeds we planted many years ago, and we see ourselves as leaders in our industry.

Will Generative AI require new skills of people working in our industry?

This technology brought a prompt-based development process. Generative AI brought this interactive mode in which you can connect with a model and get results using natural language. That’s the Big Bang - it made AI accessible to the whole world. It also created a new skill, which is Prompt Engineering. It never existed before. Through prompt engineering you can get very far and quite quickly.

So is the prompt engineer about to replace the data scientist?

Yes and no. Before GenAI, if you wanted to develop an AI based product, you needed a PhD data scientist to craft the math and magically create it for you - this is why Data Scientist was perceived as the sexiest job in the 21th century. Now, after the GenAI Big Bang, the models are already built and you basically need a prompt engineer to align the models towards your needs.

The PhDs are still needed to create the next AI technology, but they’ll be fewer and fewer of them, while today's challenges, which are mentioned above, will gradually be resolved and AI will become more and more a commodity.