Joining me to unpack it is Mushata Mattei, Partnerships lead for women in AI UAE.
Mushata, welcome to Wall Street Tamina.
First, you work on AI transformation on, um, at a major UAE bank here um in UAE.
In plain terms, what does that actually mean?
What is a bank doing differently today because of AI?
Thank you so much for having me today.
It's a pleasure.
Um, so we see, um, AI, um, is AI adoption is being accelerated massively across the region, um, and I think the easiest way to explain it is that AI transformation is not necessarily about putting a chatbot on the front of a bank.
Um, it's about looking at how the bank actually works end to end and asking where can intelligence actually make the process faster, better, or more personalized, and personalization is key.
That could mean using AI to detect fraud, to support customer service agents, help relationship managers prepare for client meetings, automate compliance processes, and improve software development to help credit teams analyze information in a better way.
Now we're already seeing this move into production.
Many banks, for example, HSBC, says it has more than 600 AI use cases in operation across areas including fraud detection, cybersecurity, transaction monitoring, and of course customer service.
More and more developers are also using coding assistance.
We've seen banks reporting almost 15% efficiency gain in coding time.
Now for me AI transformation is really about moving from where can we use AI to how should this process work differently because AI exists and from what you're seeing, Moshata, where is the real gap at the moment between AI pilots and AI that's actually running the business?
Yeah, it's a really good question, because the biggest gap isn't actually the technology.
For a long time that was the thought.
In fact, it's the operating model around the technology.
It's quite easy these days to build an impressive AI demo, and we're seeing many startups do that.
However, when it comes to pilots that can go into production, It's much harder.
It's much harder to connect a good AI demo to a bank's data systems, security compliance, and risk frameworks, and actually give people a real reason to use it.
There's some interesting evidence on that.
So Kabjam and I, for example, found that while 33% of financial institutions are developing proprietary AI agents, only about 10% have deployed them at scale.
No, I, I was saying I think the question is not can we build cool AI solutions.
We definitely can and at a much faster pace.
I think it's, it's about how can we industrialize and adopt these solutions in a responsible way.
Mushata, also there are different.
Approach when it comes to banks adopting AI, uh, banks talk about building it, building AI in house, buying it, or partnering also with tech companies which approach actually works best?
I think there's no one answer to that, and actually the most successful banks will do all three, and I've been seeing across the banking industry in the region.
Most of the banks, um, they, they, they do all three approaches, right?
So we build where the capability is strategic and gives us differentiation.
We buy where there's something already made as a mature commodity that it's easy to adopt, and we partner, uh, where we need access to frontier technology or expertise that would take years to build ourselves.
Um, the important thing is to be very clear about.
What a bank should own now customer data, risk frameworks, proprietary workflows, and the way that AI connects into your core banking environment are strategically important.
You probably don't need to build your own foundation model from scratch, but we're seeing a very hybrid approach across the industry.
So banks are partnering with major technology companies, hyper scalers, whilst simultaneously building internal AI.
Platforms and capabilities and I think partnerships become particularly important in AI because the techno technology is moving so quickly now.
I work in partnerships for banking in AI transformation, and I believe you don't need to spend 3 years building something internally when the market can already provide that solution, but you also don't want to outsource the intelligence that differentiates your bank.
Yeah, now governance, trust and human oversight, uh keep coming up in this conversation uh from where you're standing, how much decision making AI is allowed to have in banking?
It's, it's a really good question.
For me, the line should be based on the consequence of the decision rather than simply whether AI is involved.
Now, if AI is summarizing documents, helping an employee write code, or finding information, you can give it quite a lot of autonomy because the consequence of an error is relatively contained.
It's relatively small.
However, if we're talking about approving a loan, making a financial decision, Detecting potentially suspicious activity or taking an action that can materially impact a customer, I think the threshold has to be much higher.
There needs to be clear accountability, auditability, and human intervention is important where the consequences are significant.
A recent survey from Deloitte, for example, found that 94% of large banks are using generative AI, but also highlighted that there are governance gaps.
This is really the main challenge governing AI, and I don't think responsible AI means putting a human in front of every AI decision.
That's literally impossible to do at the fast pace that technology is accelerating, but it means designing a system so that the right decisions have the right level of human oversight.
Including AI, yeah, now, uh, yeah, it is one of the challenges for sure now in, in AI and banking, but also there's this really interesting point that I wanted to talk about with you today which is about women and AI and your partnerships lead for women in AI and UAE.
Is the AI talent pipeline in this region genuinely opening up for women?
Um, yes, I think it is definitely opening up, but I wouldn't say the job is, is done, and that's why organizations like Women in AI, we're a non for profit organization, and we help women enter careers within AI tech and data.
Um, I think that's why these organizations are crucial.
Um, the UAE in fact has a very unusual opportunity because it's attracting AI talent at a remarkable rate.
Um, we rated the UAE rated 2nd globally for net AI talent migration.
And AI related roles represent about 5% of job postings in the country.
I think that's magnificent.
Now the appetite for AI is definitely there.
About 75% of Middle East employees are using AI at work compared to 69% globally.
So that's already positioning the UAE in a very good place, but we have to distinguish between women using AI and women building AI.
The second question is about representation in technical roles in leadership, research, and entrepreneurship.
Women globally are still underrepresented in AI, and there's a real risk that the AI economy could reproduce some of the gender imbalance that we've seen in.
Technology, but I think the region has a lot of opportunity, but we have to be intentional about converting that opportunity into a representation.
So what, what do we still need for it actually to open up for women to be more in building AI, not just using it.
Um.
Yeah, I'm, I'm really glad you asked that question.
It's a topic that I'm very passionate about.
I would say visibility and access to the network.
We often talk about education.
Education is definitely important, but there are many highly educated women who could move into AI but simply don't see themselves represented in the rooms where opportunities are being created.
AI is developing so quickly.
There are a lot of opportunities that come through networks.
Who do you meet?
Who knows your work?
Who recommends you for something?
Who invites you into a project?
Now that's crucial, and there's a compounding effect as well.
If there aren't enough women in the room today, you have fewer women becoming mentors, investors, founders, um.
Leaders who bring the next generation through.
So for me it's not about training more women in AI.
It's about making sure women are connected to the opportunities that that training creates, the opportunities that are out there in the network, and that's exactly what we at Women in AI UAE do.
Yeah, now also when we look a year from now, um, what does AI actually running inside a major bank look like that would surprise a customer today?
Yeah, it's it's, it's a good question, and I think there are many theories on that going around.
Um, I think the biggest change will be that the customer won't necessarily know that they are interacting with AI.
I truly believe that's how good AI will be in the near future.
They will simply notice that the bank feels dramatically more responsive.
Now imagine a customer having a.
Financial need instead of navigating 5 different products, forms, and departments, an AI agent would understand the context, gather the information, recommend the appropriate next steps, and coordinate the process across the bank with the right controls, of course, and human intervention where needed.
We're already seeing the early version of this, by the way, um, we're seeing this across the US, for example, the Bank of America's EICA program has handled more than 3 billion customer interactions since its launch.
Um, HSBC has deployed as well generative AI into servicing teams supporting more than 3 million client interactions, and a lot of banks in the MINA region are embracing the same.
I think the next step is moving from AI that answers questions, which is where we are now, to AI that can actually orchestrate work, um, and that's where I think banking becomes really interesting.
It's not just a smarter chatbot, but a bank where many of the processes behind the the customer experience are intelligent, predictive, and increasingly automated.
We're actually seeing a very fast progress at the moment with AI adoption and and all sectors probably thank you so much for being with us today.
