The gap between what big tech companies say they’ll do and what they actually do has probably never been wider. This month brings six years since Google gave up its motto “Don’t be evil,” and as we’re being steamrolled into the age of artificial intelligence, the profit-chasing missteps are both laughable and scary.
The explosive emergence of the once-niche AI firm OpenAI, sparked by the public release of ChatGPT in November 2022, has knocked the tech world off its axis. Each of the world’s biggest companies has pivoted hard into the space.
For those who use their products, AI is unavoidable. Microsoft, Google and even Apple are bolting it onto tools we rely on daily, from word processors to search engines.
Remember search engines? They indexed the web so you could find the information you needed on web pages. Now, when you Google something, you’re not directed to a credible or highly cited source: A new “AI Overview” summarizes it — and keeps you on Google.
Links to the open, decentralized web are being relegated to a tab, and thanks to AI, you can ask “natural language” questions and get instant answers.
What could possibly go wrong? A lot.
The search giant has now been scrambling to clean up a flurry of bad answers, mostly from social media and satirical sites where bad answers were the point. While “hallucinations” by AI have been an innate problem of generative language models, this is AI naively taking everything on the web at face value.
How do you keep cheese from sliding off a pizza? Try glue. How many rocks should a child eat? At least one small rock a day, a Berkeley geologist recommends. Can I eat this mushroom? Yes! How do you deal with depression? Some people jump off the Golden Gate Bridge.
It’s funny, until it’s not.
Meanwhile, OpenAI got ensnared in a controversy of its own making, outright teasing that its new voice-interactive AI tool sounded like the virtual companion in the 2013 sci-fi film “Her.” The movie entity was voiced by Scarlett Johansson, who quickly lawyered up. OpenAI pulled the voices and apologized, caught doing exactly what actors and artists were worried about when they went on strike in 2023.
Billionaires seem drawn to turning sci-fi concepts into reality — and miss the underlying social commentary of such great fiction. Not to give away the ending of “Her,” but it’s not a feel-good rom-com.
The kicker? OpenAI had recently lost several of its leaders who were dedicated to “superalignment” — internal brakes that made sure the goals of the company were good for humanity. The same set of people that tried to oust CEO Sam Altman in 2023.
Google, Meta and Microsoft have also disbanded or reorganized their own AI safety teams. Yet all of these AI companies are also signing pledges and joining coalitions and government task forces focused on safe, responsible AI development.
With so much global AI hand-wringing, it can be overwhelming and frightening. One way to keep your head about you is to log off and learn and meet with people around you — and AI has unsurprisingly been the focus of many Hawaii meetings so far this year.
The Shangri La Museum recently hosted “AI in the Open,” a panel mixing local experts with industry and academia. The whole conversation is available on YouTube and worth checking out. There was the Swell AI summit in Waikiki in April focused on design, and I got to talk AI in real estate in Hilo in May.
Just yesterday, Mid-Pacific Institute hosted the AI Hawaii Summit. AI law and ethics, the future of work, AI in schools and cybersecurity filled out the agenda.
Looking ahead, Ryley Higa of AI Hawaii hosts frequent meetups on a mix of topics (including the tasty “Pie and AI”), and the Hawaii Center for AI hosts events at Hub Coworking in Kakaako.
On the education front, the state Department of Education is convening an AI in Education Summit on July 26. The DOE just released guidelines for the use of AI, which call for data protection, human oversight, fairness and equity, and accountability.
The guidelines also call for ongoing professional development, regular assessment of AI’s effectiveness in learning outcomes, and examples of school-level AI classroom use policies. It’s a solid start.
Ensuring that ethical principles are put into actual practice is clearly hard for multinational conglomerates. But we can still commit and execute, locally and individually.
Ryan Kawailani Ozawa publishes Hawaii Bulletin, an email newsletter covering local tech and innovation. Read and subscribe at HawaiiBulletin.com.
