We have been using AI for practically everything now, including finding places, teaching students, healthcare for patients, art, creative writing, the list goes on. While we love the time and cost it saves us, there is a sense that we’re losing control.
Remember when prompt engineering became a high paying job? Now AI can do prompt engineering better than humans. What happens when AI starts working with AI because that’s what’s optimum from a cost and efficiency perspective?
Read more: AI vs. human: Who is winning at being human?
As per Pascal Brier, Capgemini’s chief innovation officer, autonomous AI agents that can communicate with other AI agents will arrive by 2025. He told CNBC, a “multi-agent AI” will be a collection of agents collaborating to solve tasks in a distributed way. For example, an AI agent in marketing could run an ad campaign in Germany with another agent in legal to ensure that it’s sound.
This gives us a sense of loss of control. Are we in fact in control of AI?
As Melissa Heikkilä from The Algorithm says, no one is sure how AI actually works. “Tech companies are rushing AI-powered products to launch, despite extensive evidence that they are hard to control and often behave in unpredictable ways. This weird behavior happens because nobody knows exactly how—or why—deep learning, the fundamental technology behind today’s AI boom, works. It’s one of the biggest puzzles in AI.”
What happens when AI starts working with AI because that’s what’s optimum from a cost and efficiency perspective?
She goes on to say that the ‘intelligence’ in AI is misleading. It makes us think the tech is smart, but it isn’t, which is why it doesn’t meet our expectations.
“The technology is not truly intelligent, and calling it that subtly shifts our expectations so we treat the technology as more capable than it really is.”
We have to be aware of these flaws so that we don’t trust the tech blindly. AI might be helping generate new ideas while saving time, but it seems best to keep things casual so far. When the stakes are high, maybe human is still better, just because it’s surer.
So, does that mean AI is more hype than what it promises? But maybe the fact that it’s all hype right now is relieving. That things are more in control than we think.
Bill Dally, NVIDIA’s chief scientist testified before the US Senate, saying, “Fortunately, uncontrollable artificial general intelligence is science fiction, not reality. At its core, AI is a software program, which is limited to its training, the inputs provided to it, and nature of its output.”
Statements like these put our minds at ease. However, check out this creepy AI to AI conversation that was shared on LinkedIn. As quality of technology goes, it’s hair-raisingly good. And we know AI is getting better at a rate we hardly comprehend.
Read more: Gen AI & the news media: A deal with the devil
If you notice, towards the end of the video, the humans holding the phones begin to lose interest in the conversation, which has gone over to quantum computing and its applications. But both the bots maintain interest. Could it be that highly technological and futuristic contemplations are actually what an AI would like to talk about if they could like?
Also, the bots keep explaining because they think they’re talking to humans, while an AI app connecting to another AI app won’t need words or NLP to communicate. Given a chance, they could start communicating about things beyond our comprehension, concluding that humans are behind them or different from them in many regards.
I’m not sure I want to be around when that happens.
