Рубрика: AI

There's common agreement that generative artificial intelligence (AI) tools can help people save time and boost productivity. Yet while these technologies make it easy to run code or produce reports quickly, the backend work to build and sustain large language models (LLMs) may need more human labor than the effort saved up front. Plus, many tasks may not necessarily require the firepower of AI when standard automation will do.

That's the word from Peter Cappelli, a management professor at the University of Pennsylvania Wharton School, who spoke at a recent MIT event. On a cumulative basis, generative AI and LLMs may create more work for people than alleviate tasks. LLMs are complicated to implement, and "it turns out there are many things generative AI could do that we don't really need doing," said Cappelli.

Also: Rote automation is so last year: AI pushes more intelligence into software development

While AI is hyped as a game-changing technology, "projections from the tech side are often spectacularly wrong," he pointed out. "In fact, most of the technology forecasts about work have been wrong over time." He said the imminent wave of driverless trucks and cars, predicted in 2018, is an example of rosy projections that have yet to come true.

Broad visions of technology-driven transformation often get tripped up in the gritty details. Proponents of autonomous vehicles promoted what "driverless trucks could do, rather than what needs to be done, and what is required for clearing regulations — the insurance issues, the software issues, and all those issues." Plus, Cappelli added: "If you look at their actual work, truck drivers do lots of things other than just driving trucks, even on long-haul trucking."

A similar analogy can be drawn to using generative AI for software development and business. Programmers "spend a majority of their time doing things that don't have anything to do with computer programming," he said. "They're talking to people, they're negotiating budgets, and all that kind of stuff. Even on the programming side, not all of that is actually programming."

Also: Agile development can unlock the power of generative AI — here's how

The technological possibilities of innovation are intriguing, but the rollout tends to be slowed by realities on the ground. In the case of generative AI, any labor-saving and productivity benefits may be outweighed by the amount of backend work needed to build and sustain LLMs and algorithms.

Both generative and operational AI "generate new work," Cappelli pointed out. "People have to manage databases, they have to organize materials, they have to resolve these problems of dueling reports, validity, and those sorts of things. It's going to generate a lot of new tasks, somebody is going to have to do those."

Also: Generative AI is the technology that IT feels most pressure to exploit

He said operational AI that's been in place for some time is still a work in progress. "Machine learning with numbers has been markedly underused. Some part of this has been database management questions. It takes a lot of effort just to put the data together so you can analyze it. Data is often in different silos in different organizations, which are politically difficult and just technically difficult to put together."

Cappelli cites several issues in the move toward generative AI and LLMs that must be overcome:

• Addressing a problem/opportunity with generative AI/LLMs may be overkill — "There are lots of things that large language models can do that probably don't need doing," he stated. For example, business correspondence is seen as a use case, but most work is done through form letters and rote automation already. Add the fact that "a form letter has already been cleared by lawyers, and anything written by large language models has probably got to be seen by a lawyer. And that is not going to be any kind of a time saver."
• It will get more costly to replace rote automation with AI — "It's not so clear that large language models are going to be as cheap as they are now," Cappelli warned. "As more people use them, computer space has to go up, electricity demands alone are big. Somebody's got to pay for it."
• People are needed to validate generative AI output — Generative AI reports or outputs may be fine for relatively simple things such as emails, but for more complex reporting or undertakings, there needs to be validation that everything is accurate. "If you're going to use it for something important, you better be sure that it's right. And how are you going to know if it's right? Well, it helps to have an expert; somebody who can independently validate and knows something about the topic. To look for hallucinations or quirky outcomes, and that it is up-to-date. Some people say you could use other large language models to assess that, but it's more a reliability issue than a validity issue. We have to check it somehow, and this is not necessarily easy or cheap to do."
• Generative AI will drown us in too much and sometimes contradictory information — "Because it's pretty easy to generate reports and output, you're going to get more responses," Cappelli said. Also, an LLM may even deliver different responses for the same prompt. "This is a reliability issue — what would you do with your report? You generate one that makes your division look better, and you give that to the boss." Plus, he cautioned: "Even the people who build these models can't tell you those answers in any clear-cut way. Are we going to drown people with adjudicating the differences in these outputs?"
• People still prefer to make decisions based on gut feelings or personal preferences — This issue will be tough for machines to overcome. Organizations may invest large sums of money in building and managing LLMs for roles, such as picking job candidates, but study after study shows people tend to hire people they like, versus what the analytics conclude, said Cappelli. "Machine learning could already do that for us. If you built the model, you would find that your line managers who are already making the decisions don't want to use it. Another example of 'if you build it, they won't necessarily come.'"

Cappelli suggested the most useful generative AI application in the near term is sifting through data stores and delivering analysis to support decision-making processes. "We are washing data right now that we haven't been able to analyze ourselves," he said. "It's going to be way better at doing that than we are," he said. Along with database management, "somebody's got to worry about guardrails and data pollution issues."

Artificial Intelligence

Everyone's talking about the transformative power of emerging technologies like artificial intelligence (AI) and machine learning. That hype puts big pressure on business leaders.

Professionals are keen to start using high-profile generative AI tools, such as OpenAI's ChatGPT and Microsoft Copilot. Get the timing right for an investment in AI and your business could steal a competitive advantage. Get the timing wrong and your company could sink millions into a dead-end project.

Also: Generative AI is the technology that IT feels the most pressure to exploit

So, how do you know when to invest in emerging technologies? Four business leaders gave us their tips.

1. Focus on customer demands

Neal Silverstein, the head of technology customer services at optometry and audiology specialist Specsavers, said IT departments too often focus on "speeds and feeds" rather than making things right for customers.

Adopting emerging technology at the right time depends on understanding what your customers want, he said. "Then as long as you're validating the technology against those requirements, you'll be in the right place."

Another key consideration influencing the decision-making process is governance, especially for a company like Specsavers, which holds valuable personal data.

"There is an aspiration in our company to be more digitized in customer journeys," he said. "But there is a legal compliance that keeps us grounded. We are careful about ensuring we comply with the legal requirements for data protection."

Silverstein told ZDNET that security and governance concerns mean his company is unlikely to take a pioneering stance on AI: "We're not keen to make AI aware of colleagues' or patients' medical or financial records."

However, he said the company exploits other emerging technology, including augmented reality. Specsavers uses TeamViewer Tensor and Assist AR to remotely access and troubleshoot machines, including PCs and medical equipment.

Also: Generative AI can transform customer experiences. But only if you focus on other areas first

The technology has reduced the average handling time for each IT issue by about 15% and increased the first-contact resolution rate from 64% to 79%. This boost in operational effectiveness means staff have more time to focus on customer requirements.

"Every pair of glasses we produce is bespoke, whether that's the frame, lens, or finish the customer wants," he said. "While there are parts of a digitized journey that Specsavers will embrace and support, emerging technology must be introduced at the right level."

2. Deliver to set business outcomes

Logicalis CTO Toby Alcock is another business leader who says the key to success is focusing on whether new technology will deliver benefits, whether boosting customer experiences or increasing internal efficiencies.

"I've always looked at this question as, 'Does it add more value to our business?' If we can measure a return on investment, it's worth doing."

Also: 5 ways CIOs can manage the business demand for generative AI

Alcock said professionals must recognize some AI-led initiatives could fail to deliver a positive return. Adopt an Agile approach and test whether the technology will produce a measurable benefit.

"Dipping your toe in the water is key," he said. "Now, more than ever, we can do that with cloud services and consumption-based models. We don't have to go and buy a roomful of kit and wait six months for it to be set up."

With all that evidence in place, Alcock told ZDNET the decision on whether it's the right time to invest in emerging technology comes down to business outcomes.

"I've said my whole career, 'If this doesn't help you be more profitable, more productive, or add more value to your customers, you might as well go back to a bit of slate and a bit of a chalk because then it's just investing in technology for technology's sake.' A clear focus on business outcomes is a good starting point to measure any project."

3. Test concepts quickly

Sophie Gallay, the global data and client IT director at French retailer Etam, said knowing if it's the right time to invest in emerging technology involves a combination of elements.

Also: Agile development can unlock the power of generative AI — here's how

She said it's "super-hard" to manage all processes and priorities in parallel in businesses that aren't technology companies. Her advice to other professionals is to explore opportunities as early as possible.

"If you want to test things and prove value, I advise having teams dedicated to testing things quickly. Don't wait to put a roadmap in place to see if something has value."

Gallay recognized there's much excitement about AI. Her priority is to try and help her organization demonstrate potential benefits.

"When the markets start talking about GenAI, there is interest everywhere from the business teams. We don't have hundreds of people. I want to start putting time into something when I feel that if we validate the proof of concept, I can scale and create products."

She told ZDNET that companies face many challenges in a fast-emerging area like AI. An iterative approach can help organizations scale valuable projects rapidly.

Also: Rote automation is so last year: AI pushes more intelligence into software development

"We want to have an Agile team that is testing in a dedicated manner what generative AI can bring. Once we've proven the value, we can take the project and scale these benefits correctly, leveraging the processes of IT," she said.

"My recommendation is to use that Agile approach. If you're following standard IT procedures to test the value, you'll probably arrive at a point where the technology is already passé. There's already something new on the market that's waiting."

4. Use AI to generate ideas

Tim Lancelot, head of sales enablement at software specialist MHR, said it's important to recognize that a decision to invest in emerging technology is not a cliff edge. Smart professionals will have done their preparatory work before deciding to spend cash.

"There's nothing more frustrating than staring at a blank piece of paper and struggling with where to start," he said. "It's useful to have something you can use to generate a suggestion, provide inspiration, and save time."

As well as being the kind of emerging technology that businesses will explore and investigate, Lancelot said generative AI can also be a tool that helps professionals identify their next area of investment.

Also: 4 ways to help your organization overcome AI inertia

"The best use cases for AI are where it comes up with a suggestion and then the people, who've got years of experience and the human element, hone the idea, craft it, and give it that extra 5% or 10%."

Lancelot told ZDNET those outputs can be fed back into the system as part of a virtuous circle that produces progressively smarter suggestions. "I see AI as another team member. I don't see it as, 'The robots are coming, and they'll take our jobs.' Every tool is there to help," he said.

"And if that tool makes part of my job redundant, that's great, because I will go and find something else to do that is value-added. That success will increase the value that I can give the business and the value the business can give back to me."

Artificial Intelligence

Alibaba Group has unveiled plans to widen its cloud footprint with new data centers in key Asian markets, and has earmarked artificial intelligence (AI) as a growth driver.

"Over the next decade, no industry will be spared the disruption brought about by AI," Alibaba Chairman Joe Tsai and Alibaba CEO Eddie Wu noted in a joint letter to shareholders following the group's fiscal 2024 results. The company reported an 8% year-on-year increase in revenue to 941.17 billion yuan ($130.35 billion) for the year ended March 31, on a net income of 71.33 billion yuan ($9.88 billion), up 9% from the previous year.

Also: AI could help soften cloud sticker shock — or make things worse

With AI driving change globally, Alibaba will be looking to the technology for growth across its businesses, each of which will have use cases to offer, wrote Tsai and Wu in their shareholder memo.

Pointing to AI as "the single most powerful element that will change and accelerate the growth of our businesses," they added that AI deployment will fuel demand for computing and, in turn, growth for the company's cloud business.

"AI will not be a threat, but will herald in massive opportunities as the driver for breakthrough user experience and business models," they wrote. "If we don't keep up with the constant and marvelous improvements that AI is showing us on a daily basis, we will be displaced."

Alibaba is looking to do so through varying investment goals for its operations, including e-commerce, which is one of its two core businesses — alongside the cloud — and encompasses Taobao and Tmall Group and Alibaba International Digital Commerce Group. The latter remains nascent and needs upfront investment, while Taobao and Tmall Group is a more mature business and will need to innovate fast, according to the senior executives.

With the cloud, Alibaba is targeting to be China's leading public cloud infrastructure provider, offering a range of services spanning elastic computing, security, and AI. This focus on the public cloud will require forgoing short-term revenue from projects that yield low margins, they noted.

The Chinese tech giant, however, will be looking to expand its data center footprint globally, with new sites planned for key markets.

Also: How Alibaba's generative AI testing seeks to empower smaller e-commerce sellers

In particular, Alibaba Cloud will launch its first cloud region in Mexico and add data centers in South Korea, Malaysia, Thailand, and the Philippines over the next three years. These investments in international markets are part of Alibaba's efforts to boost its global cloud and AI offerings, the company said in a statement on Friday.

Alibaba added that its generative AI development platform, Model Studio, will be available soon for international customers via its availability zones in Singapore. This will provide access to Alibaba's large language model Tongyi Qianwen (Qwen).

Alibaba in March called off IPO plans for its logistics unit Cainiao, citing uncertain market conditions. The move followed its decision last November to backtrack on its plan to spin off its cloud business, with Wu pointing to growing US restrictions on chip sales to China as the reason for the detour.

Wu also noted the value of keeping its cloud competencies alongside the development of AI. "The deep convergence and flywheel effect of AI [plus] cloud computing will be an important impetus and advantage for our future development," he said.

Artificial Intelligence

ZDNET's key takeaways

• The Timekettle X1 Interpreter Hub is a translation device available for $700.
• The X1 Interpreter Hub has a screen and earbuds that charge when stored inside the device. Thanks to AI, it's very effective at translating and has different modes for one or two wearers per device.
• Though generally effective, the Timekettle X1 Interpreter Hub requires users to speak clearly near the device and it isn't very accurate when people speak too fast. We also can't look past the steep $700 price tag.

As a fan and proponent of artificial intelligence (AI) tools, I jump at the chance to test new, innovative applications of the technology.

That's why I decided to try out the Timekettle X1 Interpreter Hub — especially as a bilingual person.

The Timekettle X1 Interpreter Hub looks sleek and feels futuristic. It's packaged beautifully: The box contains a Timekettle (which does the translating for you), two earbuds that are stored and charged inside of the Timekettle, ear hooks and tips for the earbuds, a USB-C charging cable, and instructions.

View at Amazon

After a good charge, I turned on the Timekettle for initial testing. This is a standalone device, meaning you can translate sounds around you, like another person talking or a movie on the TV, provided it's loud enough. However, it can also handle two-way translation when each person wears an earbud. This lets you speak to a person in one language and have them hear the translation in their preferred language in their earbud and vice versa.

Furthermore, several Timekettle users can hold multilingual meetings and have up to 20 people speaking up to five languages in one place, provided each person has their own device.

The Timekettle also allows remote voice calls between two devices, as long as each is connected to Wi-Fi at the time. During these calls, each user can speak their own language and have the devices translate for the listener.

Also: My favorite XR glasses for productivity and traveling just got 3 major upgrades

I tested these functionalities and found that the Timekettle was equally effective in each instance. That is to say, it was mostly accurate, but still made mistakes, regardless of the conversation method.

As a member of a bilingual family, I tested the Timekettle X1 Interpreter Hub with my husband. We used English and Spanish one-on-one, with each of us wearing an earbud. I also tested listen-and-play mode in different languages, where one user wears both earbuds, and the device listens. Finally, I tested ask-and-go mode, which lets you speak into the Timekettle, and it displays the translation.

My intermediate proficiency in French helped me test the Timekettle in that language. Still, I used the listen-and-play mode with Korean, German, French, Spanish, and Russian with Netflix media, using subtitles to confirm the accuracy of what was generally meant.

The Timekettle X1 was accurate when using deliberately clear speech, but accuracy diminished when people spoke too fast or used regional vernacular. When online, the device can understand 93 accents in the 40 languages in its repertoire. Offline, the X1 offers 13 language pairs. The inaccurate translations were still generally understandable most of the time — though not always.

Also: Generative AI may be creating more work than it saves

I liked that the Timekettle has a clear LCD screen that displays translated text for visual confirmation, which is available in different modes. The display makes navigating and choosing the preferred translation mode easy and lets you keep track of the conversation. The visual clarity also helps with language practice, which brings me to my next point.

Aside from being a great tool for conference rooms, business conversations, international travel, and remote calls, the Timekettle X1 Interpreter Hub can also be highly useful for learning pronunciation in different languages. If you're interested in learning a new language, a device like this can greatly aid in learning how to pronounce or word a phrase correctly.

ZDNET's buying advice

Is the Timekettle X1 Interpreter Hub worth its $700 price tag? Although it's a standalone device that can be a great tool for translation, the X1, in my opinion, is priced too high for the functionality it offers. I find it makes mistakes too often to justify such a steep price. It does include earbuds and packs high-end technology that is powered by AI, so it's a definite step up from other options on the market priced between $100-150.

Though the earbuds are included, the device is incompatible with any other earbuds or headphones. You can't use the Timekettle with your AirPods or over-the-ear headphones, so at least it's a good thing that the included earbuds are comfortable. But that also means you're out of luck if you lose the Timekettle earbuds.

Also: When's the right time to invest in AI? 4 ways to help you decide

The Timekettle X1 Interpreter Hub works very well, with only minor errors, and is useful for translating in business and personal settings. It's simply priced too high for my comfort, especially when other options, like Google Translate, exist for free. However, I could see a professional interpreter appreciating this tool in their arsenal — or a well-heeled global traveler seeking a portable but reliable translation solution.

Featured reviews

Every golfer wants to be better.

It's a compulsion. Perhaps a slightly sad one.

People pay a lot of money for lessons and even more money for golf clubs in the insistent belief that they'll get close to PGA Tour standards.

Also: These 4 tech gadgets improved my every day life — and I didn't know I needed them

Still, I confess that I love to play golf. So when I saw that AI — the bliss that will soon envelop us all — had entered the golfing sphere, I simply had to see if it could turn me into a world-beater. Or, at least, get me a little closer.

Some context: I'm an OK golfer. On a good day. Those good days don't happen too often.

I prefer playing when it's extremely windy because I learned to play on courses like that. I also enjoy being outside, preferably with a good friend, and hitting the occasional shot that a professional would, ahem, envy.

I don't succumb, however, to the tech fads and fancies that many golfers do. I don't use a range finder. I don't upgrade my clubs every year as if they were iPhones. Indeed, several of my clubs predate the George W. Bush administration.

An AI golf club? Of course, I want to try it

Still, the allure of Callaway's Paradym AI Smoke collection of clubs was just too much.

So I committed some credit card points to get a Callaway Golf Paradym AI Smoke Max D Driver.

Callaway's claims for this thing are very simple. It's supposed to offer "distance and forgiveness." If you're unfamiliar with the concept of forgiveness in golf, it's the idea that even if you offer a deeply imperfect swing, the club adjusts to make your shot rather better than it deserves.

Also: 5 ways Amazon can make an AI-powered Alexa subscription worth the cost

Marketing's appeal to the human ego is never-ending, but where does the AI part come in? Well, Callaway claims this is the "world's First AI Smart Face Designed Using Real Player Data."

Essentially, like any other AI model, Callaway says it has scraped "the swing dynamics from thousands of real golfers." Those dynamics include "swing speed, club delivery, and face orientation just prior to impact."

The result? Allegedly: "Optimal launch conditions and tight downrange dispersion."

Imagine, so Callaway's promise goes, that you can have "sweet spots in the center [of the clubface] and all over the face."

That'd be like having five lovers, and they're all happy with you.

Taking my expectations to the house

Off I went, then, with my good friend Pat to the Links at Bodega Harbor, a tough course where the sea lions in the ocean make the same kinds of shrill, guttural screams as the golfers on the adjacent 16th hole.

As with all things AI, I had high expectations. I lined up for my first tee shot, believing that the ball would explode from the club face like a SpaceX rocket and soar beyond sight.

Surprisingly, this wasn't the case. I hit a house on the right side of the fairway. Nerves, you understand. Excitement, too. I'd also just got off a cross-country flight the night before.

Also: I just ordered the cheapest Surface Pro option — why I (probably) won't regret it

I feared what Pat would say. He came out with this: "AI, eh? Sweet."

Yes, I'm making excuses. Every golfer does.

Every golfer also knows you're allowed a so-called breakfast ball. It's the first tee freebie.

My second attempt went straight, but certainly no further than my previous driver. It also made a slightly soft, thwackless sound, not one I'd associate with additional distance or optimal launch conditions.

Indeed, over all 18 holes, I didn't manage to hit a single drive that gave me any more distance at all, which distressed me a little. I live with the promise that AI will make us all better and smarter, yet here I was being an entirely recognizable version of myself.

Golf, AI, and life

On the positive side, the club is very forgiving. It strains to hit the ball straight, even if your swing is to alignment what OpenAI is to consistent candor.

The driver is nicely balanced and easy to hold, and I want to believe that, in time, we'll become at one with each other and soon achieve optimal launch conditions, thanks to the thousands of people who allowed their swings to be committed to AI.

I'm a touch concerned, though, that many companies are adding AI to their wares — just those two letters — without there being any tangible benefit. It can feel a little like, dare one suggest, smoke and mirrors.

Also: Rote automation is so last year: AI pushes more intelligence into software development

My experience with the Callaway Golf Paradym AI Smoke Max D Driver — please be careful, this thing retails at just under $600 — might mirror my experience with AI in the rest of the world.

Somewhere in the margins, it might make me a little better, a little more productive in small parts of my life. Ultimately, though, I'm still myself, for better and for worse.

I'd love to philosophize a little more, but I'm off to the golf course. I need to try the Smoke again.

My first round with the AI driver may, after all, have just been a hallucination.

Bye.

Artificial Intelligence

To give AI-focused women academics and others their well-deserved — and overdue — time in the spotlight, TechCrunch has been publishing a series of interviews focused on remarkable women who’ve contributed to the AI revolution. We’re publishing these pieces throughout the year as the AI boom continues, highlighting key work that often goes unrecognized. Read more profiles here.

Miriam Vogel is the CEO of EqualAI, a nonprofit created to reduce unconscious bias in AI and promote responsible AI governance. She also serves as chair to the recently launched National AI Advisory Committee, mandated by Congress to advise President Joe Biden and the White House on AI policy, and teaches technology law and policy at Georgetown University Law Center.

Vogel previously served as associate deputy attorney general at the Justice Department, advising the attorney general and deputy attorney general on a broad range of legal, policy and operational issues. As a board member at the Responsible AI Institute and senior advisor to the Center for Democracy and Technology, Vogel’s advised White House leadership on initiatives ranging from women, economic, regulatory and food safety policy to matters of criminal justice.

Briefly, how did you get your start in AI? What attracted you to the field?

I started my career working in government, initially as a Senate intern, the summer before 11th grade. I got the policy bug and spent the next several summers working on the Hill and then the White House. My focus at that point was on civil rights, which is not the conventional path to artificial intelligence, but looking back, it makes perfect sense.

After law school, my career progressed from an entertainment attorney specializing in intellectual property to engaging civil rights and social impact work in the executive branch. I had the privilege of leading the equal pay task force while I served at the White House, and, while serving as associate deputy attorney general under former deputy attorney general Sally Yates, I led the creation and development of implicit bias training for federal law enforcement.

I was asked to lead EqualAI based on my experience as a lawyer in tech and my background in policy addressing bias and systematic harms. I was attracted to this organization because I realized AI presented the next civil rights frontier. Without vigilance, decades of progress could be undone in lines of code.

I have always been excited about the possibilities created by innovation, and I still believe AI can present amazing new opportunities for more populations to thrive — but only if we are careful at this critical juncture to ensure that more people are able to meaningfully participate in its creation and development.

How do you navigate the challenges of the male-dominated tech industry, and, by extension, the male-dominated AI industry?

I fundamentally believe that we all have a role to play in ensuring that our AI is as effective, efficient and beneficial as possible. That means making sure we do more to support women’s voices in its development (who, by the way, account for more than 85% of purchases in the U.S., and so ensuring their interests and safety is incorporated is a smart business move), as well as the voices of other underrepresented populations of various ages, regions, ethnicities and nationalities who are not sufficiently participating.

As we work toward gender parity, we must ensure more voices and perspectives are considered in order to develop AI that works for all consumers — not just AI that works for the developers.

What advice would you give to women seeking to enter the AI field?

First, it is never too late to start. Never. I encourage all grandparents to try using OpenAI’s ChatGPT, Microsoft’s Copilot or Google’s Gemini. We are all going to need to become AI-literate in order to thrive in what is to become an AI-powered economy. And that is exciting! We all have a role to play. Whether you are starting a career in AI or using AI to support your work, women should be trying out AI tools, seeing what these tools can and cannot do, seeing whether they work for them and generally become AI-savvy.

Second, responsible AI development requires more than just ethical computer scientists. Many people think that the AI field requires a computer science or some other STEM degree when, in reality, AI needs perspectives and expertise from women and men from all backgrounds. Jump in! Your voice and perspective is needed. Your engagement is crucial.

What are some of the most pressing issues facing AI as it evolves?

First, we need greater AI literacy. We are “AI net-positive” at EqualAI, meaning we think AI is going to provide unprecedented opportunities for our economy and improve our daily lives — but only if these opportunities are equally available and beneficial for a greater cross-section of our population. We need our current workforce, next generation, our grandparents — all of us — to be equipped with the knowledge and skills to benefit from AI.

Second, we must develop standardized measures and metrics to evaluate AI systems. Standardized evaluations will be crucial to building trust in our AI systems and allowing consumers, regulators and downstream users to understand the limits of the AI systems they are engaging with and determine whether that system is worthy of our trust. Understanding who a system is built to serve and the envisioned use cases will help us answer the key question: For whom could this fail?

What are some issues AI users should be aware of?

Artificial intelligence is just that: artificial. It is built by humans to “mimic” human cognition and empower humans in their pursuits. We must maintain the proper amount of skepticism and engage in due diligence when using this technology to ensure that we are placing our faith in systems that deserve our trust. AI can augment — but not replace — humanity.

We must remain clear-eyed on the fact that AI consists of two main ingredients: algorithms (created by humans) and data (reflecting human conversations and interactions). As a result, AI reflects and adapts our human flaws. Bias and harms can embed throughout the AI lifecycle, whether through the algorithms written by humans or through the data that is a snapshot of human lives. However, every human touchpoint is an opportunity to identify and mitigate the potential harm.

Because one can only imagine as broadly as their own experience allows and AI programs are limited by the constructs under which they are built, the more people with varied perspectives and experiences on a team, the more likely they are to catch biases and other safety concerns embedded in their AI.

What is the best way to responsibly build AI?

Building AI that is worthy of our trust is all of our responsibility. We can’t expect someone else to do it for us. We must start by asking three basic questions: (1) For whom is this AI system built (2), what were the envisioned use cases and (3) for whom can this fail? Even with these questions in mind, there will inevitably be pitfalls. In order to mitigate against these risks, designers, developers and deployers must follow best practices.

At EqualAI, we promote good “AI hygiene,” which involves planning your framework and ensuring accountability, standardizing testing, documentation and routine auditing. We also recently published a guide to designing and operationalizing a responsible AI governance framework, which delineates the values, principles and framework for implementing AI responsibly at an organization. The paper serves as a resource for organizations of any size, sector or maturity in the midst of adopting, developing, using and implementing AI systems with an internal and public commitment to do so responsibly.

How can investors better push for responsible AI?

Investors have an outsized role in ensuring our AI is safe, effective and responsible. Investors can ensure the companies seeking funding are aware of and thinking about mitigating potential harms and liabilities in their AI systems. Even asking the question, “How have you instituted AI governance practices?” is a meaningful first step in ensuring better outcomes.

This effort is not just good for the public good; it is also in the best interest of investors who will want to ensure the companies they are invested in and affiliated with are not associated with bad headlines or encumbered by litigation. Trust is one of the few non-negotiables for a company’s success, and a commitment to responsible AI governance is the best way to build and sustain public trust. Robust and trustworthy AI makes good business sense.