There’s a lot of mystery and glamour surrounding the future of artificial intelligence, which suddenly looks to be gaining capabilities at an incredible pace. Yet the nonstop product announcements now coming from Big Tech have largely been centered around quotidian office tasks like making spreadsheets or managing Gmail or creating PowerPoint slides.
While that’s disappointing on some level, it’s a perfectly reasonable business strategy.
Microsoft, recently, integrated OpenAI’s technology into its office tools like Word and Excel. Flashy it’s not. But by staying in their domain, tech’s giants can remain competitive while quickly releasing products to the public.
Software companies presumably could troubleshoot those kinds of issues easily when it comes to building a slide deck or summarizing a business meeting. When it comes to higher-stakes applications, the technical problems become more difficult to solve.
“I worry most about mistakes in high-stakes situations, like legal or medical diagnoses,” said Graham Neubig, a computer scientist at Carnegie Mellon. “Humans make mistakes in those cases, too,” he added.
When OpenAI released ChatGPT-4 earlier this week, the startup said additional capabilities could lead to new risks, suggesting that more problems will come with pushing the limits of AI.
If the upshot is that big tech will be conservative in how it unleashes AI to the masses, it seems a sensible tradeoff.
The current work-related applications of AI are still useful, even with the occasional bug. Google Translate was introduced in 2006 and it still makes mistakes—sometimes hilariously so. That doesn’t mean it isn’t useful, said Neubig.
Meanwhile, the next batch of AI integrations in the workplace, if not all that exciting, promises to at least be helpful for productivity. “The technology is pretty impressive already,” said Neubig.