The majority of AI and machine learning is built to bolster the revenue streams of corporations rather than for the interest of human welfare.
That scares me.
The source of this fear comes from three compounding causes:
- Lack of effective government investment in AI
- Venture backed investors favoring business to business (B2B) startups over business to consumer (B2C)
- Corporations primarily driven by profit, not social utility.
Our society lacks a singular force with deep pockets that consistently values social good over revenue. The closest should be government, but it’s sluggish at best and apathetic at worst when it comes to technology adoption that benefits society. Furthermore, the United States government is primarily fueled by the pockets of lobbyists who represent the interest of corporations rather than citizens. Occasionally there’s a bipartisan initiative to utilize technology to empower citizens but even then the implementation of such initiatives waxes and wanes depending on election and budget cycles.
The bottom line is that government is generally slower at technology adoption than corporations, which are more efficient at utilizing new technology under the pressure of competition. Case in point, Amazon won’t stop recommending a shiny KitchenAid blender after I asked Alexa for a cake recipe.
Another major factor is how startups are created and nurtured in today’s market. 75% of venture investment dollars across all fundraising stages go towards B2B startups for the past decade [MoneyTree]. The benefits of serving B2B over B2C are concrete for burgeoning companies, as businesses have well-defined needs and are willing to pay for services that benefit their bottom line. People, on the other hand, have more diversified needs and are used to getting software services for free. Many B2C startups pivot to B2B after failure to gain user traction or seed investment. This has been particularly true in HR tech, as we’ve seen handfuls of machine learning and AI powered startups pivot their target market from jobseekers to businesses.
Lastly, mature companies’ interests are misaligned with humanity. Social utility always comes second to profit maximization, which is to no fault to the corporation – they are, by design, accountable to their shareholders. An optimist may argue that companies only build products or deliver services that bear value to consumers, that is, it’s not a zero sum game between companies and consumers. That doesn’t mean consumers will end up winning, either. Did I really need that KitchenAid blender? Or was it brilliant machine-learning backed marketing on Amazon’s part?
These compounding factors may seem bleak for humanity without a drastic change in the structural fabric of our current society. On the other hand, mingled with that fear of futuristic AI and malevolent machines, is a healthy dose of hope. Conditions aren’t optimal, but there is room for successful products with high social utility driven by human need.
For one, there are a lot of initiatives of ‘AI for Good.’ Some of these initiatives are backed by wealthy philanthropists and corporations with a conscience and others are powered by unified governments around the world. These nonprofits and NGOs demonstrate there is a strong desire and growing recognition that society needs to utilize AI for the greater good.
Second, the high technology barrier and time needed to implement machine learning has drastically decreased with the availability of Machine Learning as a Service (MLaaS) platforms. Google Cloud Predict API, Microsoft Azure Machine Learning, AWS SageMaker, and Anaconda are major players in lowering the cost for anyone to dabble in basic machine learning.
Third, consumers do have a choice and voice in determining the trajectory of technology creation. The amalgamation of our pocketbooks has the power to hold corporations accountable for social irresponsibility as well as reward them for good deeds. In fact, this shift towards socially responsible purchases is currently happening, as Berkeley in a recent study found that “more than 9-in-10 millennials would switch brands to one associated with a cause.”
Regardless, even given these positive forces pushing AI for good, we’re still missing a major piece to tip the scale in favor of AI for humanity rather than corporations. There’s a stark absence of compelling incentives to encourage, incubate, and sustain machine learning startups driven by social good.
To fill this void, we can consider a few ideas. One is for philanthropic grants to subsidize sustainable machine learning startups. Another is for government to provide continual incentives (financial and regulatory) to not only help social impact AI startups germinate but also foster long-term viability. For some of the toughest social issues that can benefit from machine learning, federal or local government can create a bid system for companies (or individuals) to crowdsource an intelligent solution with reward. Taking it a step further, government could set social responsibility benchmarks for corporations, particularly concerning responsible use of nascent technology in this age of nebulous personal data rights.
I’m the CTO of Jobscan, a service that uses machine learning to help jobseekers with their resumes and professional profiles. Our vision is to create a service that will eliminate the stress of unemployment – where one day, our AI will deliver your perfect job offer, logistically, culturally, professionally with zero down time. We’re lucky and grateful for where we are, but the path forward to continually invest in AI won’t be easy. Without a refactor of our current corporate constructs to incentivize social good over profit, it’s an uphill battle for any company that chooses social good to survive or thrive.
About the Author