Responsible synthetic intelligence (AI) must be ingrained in the DNA of a company.
“What is it about bias in AI that we all need to think about today?” “It’s because AI is powering everything we do today,” Miriam Vogel, president and CEO of EqualAI, told a live broadcast audience at this week’s Transform 2022 event.
Vogel discussed AI bias and responsible AI in depth during a fireside talk moderated by Victoria Espinel of the trade group The Software Alliance.
Vogel has extensive knowledge in expertise and policy, having worked in the White House, the US Department of Justice (DOJ), and the NGO EqualAI, which is dedicated to reducing unconscious bias in AI development and application.
She is also the head of the recently formed National AI Advisory Committee (NAIAC), which was created by Congress to advise the President and the White House on AI policy.
As she stated, AI is becoming increasingly important in our daily lives — and vastly improving them — but, at the same time, we must be aware of the differences and inherent hazards of AI. Everyone — builders, producers, and buyers alike — should consider AI to be “our partner,” as well as environmentally clean, efficient, and dependable.
“You can’t establish confidence in your app if you’re not confident it’s secure for you and made for you,” Vogel explained.
Now is the time
We must address the issue of responsible AI now, according to Vogel, since we are still building “the rules of the road.” What characterises AI remains a “grey area.”
And what if it’s not addressed? The consequences might be severe. As a result of AI prejudice, people are unlikely to receive adequate healthcare or employment opportunities, and “litigation will come, regulation will come,” said Vogel.
When this happens, “we can’t disentangle the AI systems on which we’ve become so reliant, and which have gotten entangled,” she says. “Right now, now, is the moment for us to be very attentive about what we’re constructing and deploying, ensuring that we’re analysing risks and decreasing them.”
Good ‘AI hygiene’
Companies should address responsible AI immediately by developing strong governance processes and regulations, as well as fostering a safe, collaborative, and visible culture. Vogel stated that this must be “placed through the levers” and dealt with thoughtfully and purposefully.
In recruiting, for example, companies might begin by simply inquiring whether or not platforms have been tested for prejudice.
“Just that fundamental query is really powerful,” Vogel said.
An organization’s HR team must be backed with AI that is inclusive and does not exclude the most effective applicants from employment or advancement.
It’s an issue of “proper AI hygiene,” according to Vogel, and it starts with the CEO.
“Why is the C-suite necessary?” “At the end of the day, if you don’t have buy-in at the highest levels, you can’t get the governance framework in place, you can’t have investment in the governance framework, and you can’t get buy-in to verify that you’re doing it correctly,” Vogel explained.
Furthermore, bias identification is a continuous process: once a framework has been developed, a long-term course of must be in place to regularly monitor whether or not bias is hampering programmes.
“Bias may embed at every human touchpoint,” according to Vogel, from data collection to testing, design, development, and deployment.
Responsible AI: A human-level downside
Vogel said that the discussion of AI bias and AI obligation was previously limited to programmers — but Vogel believes this is “unfair.”
“We can’t expect them to fix humanity’s issues on their own,” she said.
It’s human nature for people to think only as broadly as their knowledge or imagination allows. So, the more voices that can be brought in to learn best practices and ensure that the age-old problem of prejudice does not permeate AI, the better.
This is already happening, with countries all over the world developing regulatory frameworks, according to Vogel. As an example, the EU is developing a GDPR-style law for AI. Furthermore, in the United States, the Equal Employment Opportunity Commission and the Department of Justice recently issued a “unprecedented” joint declaration on reducing disability discrimination – something AI and its algorithms may exacerbate if not monitored. The National Institute of Standards and Technology was also ordered by Congress to develop an AI risk management framework.
“We should anticipate a lot from the United States in terms of AI legislation,” Vogel said.
This comprises of the newly formed committee, which she currently chairs.
“We’re going to make a difference,” she said.
The metaverse is a $800 billion market, according to a recent Bloomberg Intelligence analysis. Others disagree on what the metaverse is, but with so much money and interest surrounding it, it has everyone talking.
Without a question, AI will play a significant role in the metaverse, particularly in how we connect with others. While humans will be more linked than ever before, AI that is not bound by any government, standard, or ethical code might have disastrous consequences. “Who gets to make the rules?” former Google CEO Eric Schmidt recently wondered.
Understanding the implications of AI
Because AI algorithms are designed by individuals with prejudices, they can be programmed to replicate their creators’ thinking patterns and biases. We’ve seen how AI can produce gender prejudice, or how AI can offer males bigger credit card limits than women, or how particular races are more susceptible to unfair bias.
Dark AI behaviours that might produce and perpetuate prejudice must be addressed in order to establish a flourishing and more equal metaverse. But who gets to make the call? And how can people avoid being biassed?
The solution to “unchecked AI” is to establish ethical norms across all enterprises. Dark AI patterns, in our opinion, can be intrusive. The majority of AI is produced without ethical control, which must change in the metaverse.
We need to be sure that language AI is trained to be ethical as well.
Using AI to translate messages in the metaverse
As an active language learner and the creator of a company that utilises AI and humans to connect people around the world, I’m delighted about the potential of everyone becoming great polyglots — able to speak many languages — but I’m even more curious about how that AI will operate.
Many users in the metaverse will most likely communicate in their native languages, with potential AI-based language translators. If we are not careful, AI-powered language technologies can perpetuate bigotry. We must ensure that language AI is also ethically taught.
Assume Joe’s avatar wishes to communicate with Miguel’s avatar, but Joe and Miguel do not share a common language. How does AI interpret their messages? Directly? Or should we interpret for the intent of the person rather than literally so that the person receiving the message understands?
Blurring the lines between human and machine
It will matter how “human” we are in the metaverse. Businesses may employ language technology to instantly translate conversations into several languages, fostering online community, trust, and inclusiveness.
However, if we are not cautious with the language we use, technology might promote bias or enable uncivil conduct. How so? Have you ever heard a three-year-old converse with Alexa? Personable is not the right term. People do not feel the need to be nice when they know they are communicating with technology rather than genuine humans. Customers are instead disrespectful to chatbots, Amazon’s Alexa, and automated phone lines. The list goes on and on.
In an ideal future, AI for a language would capture the subtleties and empathy required to authentically depict a person, transforming the metaverse into a place where humans and technology may coexist.
Impersonal AI in the metaverse might be harmful as well. The correct language may foster genuine emotional connection and comprehension. The appropriate messaging may help humanise a business via AI-powered language operations.
Technology that allows brands to communicate in several languages immediately will be critical. We feel that native language builds consumer trust. But how can a virtual community with no borders have a native language? And how can such an atmosphere foster trust?
As previously said, the metaverse offers enormous potential for firms seeking increased visibility in a virtual environment. People are already paying big bucks for virtual fashion, and this trend is only going to become worse. Brands must figure out how to create online experiences that seem real, if not superior, to in-person interactions. That is a high bar to clear, and clever language communication will play a role in getting there.
Nobody knows what the metaverse will look like in the end. However, no one wants to be remembered as the brand that disproportionately harmed one group of people over another or dehumanised their product. AI will become increasingly adept at anticipating trends for the better. However, if AI is not checked, it might have major consequences for how humans “life” in the metaverse. That is why responsible or ethical AI ethics are required.
There are various possibilities to lose client trust or human sentiments when AI drives language, chatbots, or businesses’ virtual worlds. It is up to AI academics and specialists to collaborate with companies to develop solutions for ethical AI frameworks so that we can “live” happily in the metaverse.