AI has been getting a bad rap when it comes to possible bias. Amazon’s AI recruiting engine, ironically developed to eliminate possible human prejudice, appeared to be biassed against women. It downgraded resumés that contained words like “women” and the graduates of two all-women colleges.  Google Photos’ algorithm apparently showed bias against black people, labelling the photos of a black software developer and his friend as gorillas. And a face recognition algorithm used by police forces across the globe, including processing 30 million face shots in the USA, made significant mistakes in identifying black women as compared to white women or even both black and white men.

Big brains around the world are working day and night to come up with maths models to explain what’s really happening inside AIs. AI ethics conferences are taking place. Papers are being written. And progress is being made. Fujitsu Laboratories, to take one example, has developed a sophisticated mathematical model it calls Wide Learning, to understand what is happening inside “black box” AIs. 

A more neutral way to evaluate loan applications 

For all the concern about AI bias, there is another side to the story. When it comes to new business financing, AI can actually reduce the prejudice often encountered by entrepreneurs who don’t fit the mould.

Most traditional banks still operate largely manual systems, where an employee reviews a loan application made up of a description of the business, profiles of the founder and any key executives, a business plan, some key milestones, and a cash flow forecast.

That all seems perfectly logical, but everyone has their preferences and expectations — no matter how good a bank’s unconscious bias training has been. 

The first issue is that banks just tend to be biassed against SMEs. Tobias Baer, an independent advisor and scholar cooperating with the University of Cambridge in data science, risk management, and psychology, says that bias against SMEs is systemic. “It is practically embedded in the products, technology and processes that have remained largely unchanged for the past 30-plus years,” he says.

There are three levels where systemic bias against SMEs comes into play. There is strategic bias, which selects against SMEs as they are perceived to be risky, expensive and difficult to serve. Decision bias imposes the same criteria to decide whether to lend to a large business or a small one. And cognitive bias creates a natural preference to work on fewer larger loans rather than putting more aggregate effort into many smaller ones. 

Essentially, the lending industry is engineered around lending to large companies, with 57% of SME applications for credit abandoned or rejected.

Bias against diversity

But the bias issue goes even deeper than that. The New York Times recently reported on high levels of racial bias in the US government’s $800 billion Paycheck Protection Program during the pandemic, when minority entrepreneurs, especially Black business owners, struggled more than white borrowers to find a willing lender. 

A research project at New York University’s Stern School of Business found that the problem was particularly pronounced at smaller banks — with human bias as the main reason. 

When technology was making the lending decision, the bias issue evaporated. The majority of Black borrowers who received aid from the program received it from a financial technology company, not a bank. The automated loan vetting and processing systems used by fintechs significantly improved approval rates for Black borrowers, demonstrating how technology can help level the playing field.

Gender bias is in play too. A 2016 study showed that female-owned businesses receive loan approvals 33% less often than male-owned businesses.

Bias exists at VCs as well as banks

VCs are as prone to being biassed and blinkered as banks, when it comes to choosing investments.

For a start, the industry is not very diverse. A Deloitte study based on data from over 200 US firms representing over 2,700 employees, reports that just 3% of US venture capitalists identified as Black or Latino in 2018, the most recent year for which data is available. 

Quite possibly related to that, just 1% of US venture capital goes to Black-owned startups, according to data from the venture-tracking firm Crunchbase.

And gender bias is strong in VCs too. Again, the landscape lacks diversity, including in Europe, where senior positions are heavily male. A 2021 report by the British Venture Capital Association (BVCA) found that just 10% of senior investment roles in Europe were held by women. Women from ethnically diverse backgrounds represented only 9% across all grades and functions. Within senior roles, only 3% of employees were women from ethnically diverse backgrounds, with Black women accounting for fewer than 1% of senior roles.

This tends to play out in various negative ways for women seeking VC funding. The FT’s Sifted recently carried a horror-story catalogue of gender bias experiences. A female first-time founder was described as “silly” by a male VC when she expressed a perfectly reasonable preference for a VC partner with at least one female decision maker. The male VC retorted he wouldn’t be “lectured on female empowerment” because he “has four daughters”.

Fintech is reducing bias

Despite these cringe-making examples, there is hope. Diversity in banks and VCs is improving. The BVCA’s 2021 report also concludes that the number of women working in investment and non-investment roles has improved. All male investment teams have declined to 12% down from 28% in the 2018 report.

And, as the NYU Stern School of Business research demonstrates, the rise of fintechs is turning bias around too. AIs are not perfect, but the sort of unconscious biases based on education, skin colour, religion, name or gender can be programmed out. Fintechs’ decisions are therefore much more likely to be based purely on the metrics and data, without these built-in biases.

The problem is, the conventional sources of financing have been slow to adopt the technology. We are well down the fintech revolution path now. My own company has been in operation for six years. There are others like us. But the big players have been remarkably slow to wake up to new ways of working with AI. Two major names, NatWest and Warburg Pincus, have taken until May this year to announce that they are just “contemplating” the launch of an online lender for small businesses.

VCs are possibly even greater tech-laggards. “As an ex-VC,” says a fintech CE quoted in a recent article, “I can say that often the most sophisticated technology they use is a CRM…I think VCs want to be much more data-driven, less biassed and able to look at companies in a more structured way, but it’s hard to do it without technology.”

Why bias against SMEs matters

Debt financing is hugely important for SMEs. The EU’s latest Survey on the access to

finance of enterprises (SAFE), for 2021, finds that in all EU27 Member States, 77% of SMEs used debt financing in some form in 2021.

However, they are typically overlooked by the traditional banks. The same SAFE report finds that the difficulty of accessing business finance increases significantly as business size decreases.

Obtaining debt finance can be transformative for SMEs. A February 2022 report from the European Central Bank (ECB) on the impact of fintech lending to SMEs, with an emphasis on Peer-to-Business (P2B) lending, notes firms that achieve funding grow faster, with “an 8.2 percentage points increase in asset growth, a 5.1 percentage points increase in employment growth, and a 5.8 percentage points increase in sales growth relative to the control group of rejected applicants.”

The ECB also notes that larger banks tend to vacate SME lending when the financial system is under stress, or raise interest rates and impose other unfavourable terms. The data shows that it is non-traditional lenders, including fintechs, that fill the void, offering unsecured loans on more favourable terms. 

As the ECB concludes, the role of fintechs in promoting SME growth is proving to be vital for this important segment of the business community. There are many reasons why fintechs are proving so attractive to SMEs. In my opinion, the use of AI by fintechs is an important factor in the substantial decrease in bias against SMEs and the diverse entrepreneurs who run these businesses.