The Week in AI: Nuking the AI genie & run GPT locally on your laptop
When it comes to AI, at this point:
The genie is out of the bottle.
The horse has bolted from the stables.
The cat is out of the bag.
The toothpaste is out of the tube.
The ship has sailed.
(Thanks, GPT-4, for the help with idioms).
Now that AI and LLMs are here, you can’t turn back the clocks (sorry, I’ll stop).
And yet, the calls for stopping the train that has left the station (okay, seriously, I’ll stop now) are reaching a fever pitch this past week:
The open letter in question urgently calls for a “pause in giant AI experiments,” saying:
But if you pay attention, you’ll notice one thing:
“Should we let machines flood our information channels with propaganda and untruth?”
Signatories like Yoshua Bengio, Stuart Russell, Elon Musk, Steve Wozniak, Andrew Yang, and especially Yuval Noah Harari did not suddenly become altruistic and concerned about your free speech, equal opportunity, or equal opportunity well-being.
This is, fundamentally, about who gets to control AI and ML—and, therefore, control what the rest of us can say and do.
Never mind that pausing, stopping, or outright banning ChatGPT or any other tool is practically impossible and unenforceable.
Nevertheless, Italy is trying:
But, let’s be honest, the only way to get close to succeeding is if you go scorched earth, like Eliezer Yudkowsky demands:
He has a point. When you come for the king, you better not miss.
Or, as GPT-4 helpfully suggests:
Swing for the fences.
In for a penny, in for a pound.
Don't pull your punches.
No half measures.
The irony is that it’s probably only possible to enforce a pause, ban, or restriction of AI with AI.
As we’ve also seen this week, total surveillance and control are what they’re hoping to do under the guise of a TikTok ban.
Run models on your computer and open-source alternatives
Perhaps one viable path forward in using AI and Large Language Models with less (or no) bias is by running your own open-source bot, on your computer:
Good news on the open source front this week:
“At Cerebras, we believe democratizing large models requires both solving the training infrastructure challenge and opening more models to the community. To that end, we have designed the Cerebras Wafer-Scale Cluster with push-button scaling, and we are open-sourcing the Cerebras-GPT family of large generative models.”
Now, where have we heard of “democratizing large models” before? Ah, yes, OpenAI. We’ve all seen how that turned out. Still, this could become beneficial.
In other AI news around the internet
Interesting things have been happening, not just in the past week but recently:
Robots taking our jobs: Generative AI set to affect 300 million jobs across major economies.
They’re starting small: Agility’s latest digit robot prepares for its first job: moving totes.
But soon videographers are done: Runway Gen-2 is the First Publicly Available Text-to-Video Generator.
Screenwriters see the writing on the wall and are coming to terms with it: WGA Would Allow Artificial Intelligence in Scriptwriting, as Long as Writers Maintain Credit.
Then again, if you put monkeys in a room with typewriters, will they write a masterpiece? Who knows. But this monkey has an implanted Neuralink brain chip and is playing a game of pong with his brain.
Until next week, keep your eyes peeled, your ears open, on your toes, your fingers on the pulse, and your head on a swivel (sorry, GPT-4 made me do it).
RETURN is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.