Does AGI Really Matter or Just More Hype?
Does AGI Really Matter or Just More Hype?
Silcon Valley revs up its latest marketing message
Artificial general intelligence or AGI will arrive within five years, says AI industry marketing engines at NVIDIA, OpenAI, and others. Should enterprise executives care, or is this another hype distraction?
The TLDR version: Pass, AGI is hype and won’t impact the enterprise anytime soon.
So what gives?
Redefining AGI for the Shine
AGI or as it was originally defined, sentience in a machine, is actually the primary source of fear about AI in the public dialogue. Hollywood tropes and dramatic journalistic conjecture debate whether brilliant and sometimes malevolent AIs could take over the world and replace humans at every corner.
Like all things, marketing hype is now redefining and watering down the definition to build something more sellable. Today, AGI is in the midst of repositioning to mean “software that performs tasks it is not necessarily trained or developed to perform.” Or an even more watered-down version where it just performs tasks well.
Consider how the definition of industry leader OpenAI has changed. , “In February 2023, OpenAI said, “Our mission is to ensure that artificial general intelligence—AI systems that are generally smarter than humans—benefits all of humanity.”
In January 2024, we have a summary from a Sam Altman Davos talk: “AGI is a term used to refer to a form of artificial intelligence that can complete tasks to the same level, or a step above, humans.”
The move to AGI is fueled by a need to have a new sheen on AI technologies after the general sense of deflation with the generative movement in the enterprise. Notice the current marketing version of AGI promises tasks performed as well as humans compared to today’s generative AI technologies, it definitely needs governance guardrails and close human supervision.
Is Hype a G Thing
God knows it wouldn’t be the first time a technology has turned to a G thing to respin their promising technology (not me or partner Greg).
In the late 1990s Personal Communications Services or (PCS) technology promised to deliver the Internet to cellular phones. The technology rolled out, and customers integrated paging technology onto their phones. Web access was painfully slow, returning users to bad memories of 9600 baud modems. While the paging industry was wiped out as a result, PCS enjoyed limited success.
Along comes 3G, promising “broadband” Internet to smartphones delivering the web and apps and so much more. Launched in 2001, 3G rolled out and it was a massive success. The strain of the iPhone-fueled app revolution brought a need for 4G, and video helped fuel the need for today’s 5G networks.
One could argue that 3G was the harbinger of the end of landline communications and the unplugged movement. For the first time, consumers were untethered from landlines and PCs, allowing them to communicate wherever they were however they wanted.
Will a G help AI succeed?
Unlike 3G, AI’s problems are deeper than provisioning bandwidth. In this case, if provisioning computing power was the only big challenge, then NVIDIA’s claim that AGI would come in five years would be reason for pure investor glee.
Unfortunately, based on current performance, the software still has significant breakthroughs ahead of time to achieve AGI, and those are just tasks. Let’s not even entertain sentient machines (although fear-mongering AI journalists certainly will). Moving forward, will Claude 5 or GPT 7 powered by nextgen NVIDIA chips achieve that vision? It’s likely, but when is another matter.
One thing is certain: AGI’s arrival will be declared before it is actually achieved. Will the declaration spawn another wave of excitement, disappointment, or a combination of both? Probably both.
Expect less tolerance for over-fueled promises this time, though.
All images created with Midjourney Alpha.
