Elon Musk, the person with extra outrageous ventures than fingers, is at it once more. This time, he’s saying a brand new startup known as TruthGPT, which he claims would be the final truth-seeker, “making an attempt to grasp the character of the universe.”
But let’s be actual right here. Is Musk actually making an attempt to avoid wasting us from the looming risks of synthetic basic intelligence (which isn’t even a factor but), or is that this simply one other PR stunt to spice up his ego?
Musk vs. ChatGPT: The Battle of the AIs
The world wasn’t prepared for ChatGPT when OpenAI first unveiled it in December 2022. With just some easy prompts, ChatGPT can generate complete articles, essays, jokes, and even poetry. It’s not excellent, nevertheless it’s eerily good.
By January, it had already reached 100 million lively customers. For comparability, it took TikTook 9 months and Instagram two and a half years that they took to succeed in the identical variety of customers. Since the arrival of the web, no app or on-line service has been in a position to obtain such a quick launch ramp.
The chatbot’s ‘reasoning’ talents are so good that ChatGPT-4 (the most recent public model) can rating among the many prime 10 p.c of scholars on the Uniform Bar Examination, which suggests ChatGPT can theoretically qualify as a lawyer in 41 US states and territories. It may rating 1,300 out of 1,600 on the SAT and a 5 out of 5 on Advanced Placement highschool exams in biology, calculus, psychology, statistics, and historical past.
But whereas ChatGPT is extremely human-like in its replies and interactions — virtually frighteningly so — it’s removed from excellent. Its greatest flaw is that it nonetheless makes stuff up.
All present chatbots, whether or not they’re made by the Microsoft-backed OpenAI or Google, undergo from this downside, which AI researchers affectionately name “hallucinating”. Depending on what you ask it to do, the chatbot will virtually at all times comply and make up stuff on the fly if it has to. When pressed for sources, it can invent faux web sites, science journals and papers, and even fictitious individuals simply to show its level — all with the utmost confidence.
In some situations, these complicated and inaccurate statements churned out by ChatGPT can cancel out its robust factors. ChatGPT will be extremely insightful and provoking throughout one second however dumb as a brick the following.
This is a extremely powerful nut to crack as a result of such programs shouldn’t have actual reasoning talents the way in which individuals do, so it is vitally difficult to make them perceive what’s true and what isn’t.
And that is the place Musk is available in. Or claims to return in.
TruthGPT on Fox News: Oh, the irony
Musk’s TruthGPT is being touted as the reply to ChatGPT’s flaws. According to Musk, TruthGPT can be a “most truth-seeking AI,” versus ChatGPT’s incoherent ramblings.
Okay, positive, Elon. But let’s not neglect that you just have been squeezed out of OpenAI, the corporate behind ChatGPT, after you have been initially a founding board member in 2015 and you tried to take the company over (and failed). Could this be a case of bitter grapes, or is there actually a necessity for a brand new AI language mannequin?
If you thought issues couldn’t get any weirder, Musk made this announcement on Fox News throughout an interview with Tucker Carlson. Yes, the identical Tucker Carlson who’s the embodiment of fake news. The man’s job is to literally lie all day on nationwide tv.
The irony is palpable. Musk is selling a truth-seeking AI on a community that’s notorious for its lack of factual reporting. My head is spinning. But hey, not less than the identify “TruthGPT” has a pleasant ring to it, proper?
“I’m going to begin one thing which I name TruthGPT or a most truth-seeking AI that tries to grasp the character of the universe,” Musk mentioned. “And I feel this is likely to be the perfect path to security within the sense that an AI that cares about understanding the universe is unlikely to annihilate people as a result of we’re an attention-grabbing a part of the universe.”
If you’re scratching your head proper now, you’re not alone.
Fox News is the proper viewers to drop doomsday situations as a result of the viewers are fed fear-mongering on a moment-by-moment foundation. That’s the community’s enterprise. I imply, take a look at Tucker’s face in the course of the interview.
Like any grifter value their salt, Musk reveals an existential downside you by no means thought you had prior, after which instantly affords the answer. His resolution.
But why? To save humanity after all.
In Musk’s imaginative and prescient, a brilliant basic synthetic intelligence could be able to dooming us all, however with the best constraints pioneered by TruthGPT, it might use this energy responsibly. Musk affords a daft instance as if as an example this level.
“We acknowledge humanity may resolve to search out all of the chimpanzees and kill them. We’re truly glad that they exist, and we aspire to guard their habitats,” mentioned Musk, whose different firm Neuralink has attracted the fury of animal welfare teams for its cruel experiments on monkeys which might be implanted with gadgets of their brains.
And it was simply final month that Elon penned an open letter urging AI firms and labs to pump the brakes on AI growth for six months, claiming that current advances pose “profound dangers to society and humanity.”
Why six months although? Is that how a lot time it’s good to launch TruthGPT?
How does a truth-seeking AI work, anyway?
But let’s put the irony apart for a second and concentrate on the true query: how does a truth-seeking AI truly work?
According to Musk, TruthGPT can be skilled on a corpus of “truthful” knowledge and use this data to generate responses to queries. But right here’s the factor: what counts as “truthful” knowledge?
Truth is a difficult factor. It’s not at all times black and white. What one particular person considers to be true may not be true for another person.
But that’s what science is for, you would possibly quip — and I agree. But are we going to fake that TruthGPT can be utilized by individuals solely to ask about how moons Jupiter has or what’s a black gap? As a generative AI, TruthGPT can be used for all sorts of inquiries, together with people who shouldn’t have a single proper reply or which can require educational coaching as a way to totally internalize the nuances of a thorny subject.
This is all of the extra complicated contemplating Musk had a really controversial interview with the BBC last week. When pressed by the BBC journalist on whether or not his tumbling dealing with of Twitter is selling an increase in misinformation and hate speech on the social media platform, Musk instantly retorted “Who needs to be the arbiter of fact?”
He then went on to query whether or not journalists are honest arbiters of fact, including he had extra belief in “extraordinary individuals” and at last calling the BBC journalist a liar as a result of he couldn’t give you examples of Twitter’s mismanagement of hate speech on the spot.
With this newest announcement, issues are coming collectively. It is quite clear that Musk has been planning this transfer into generative AI for months now and his dance round ‘what’s fact’ has a particular purpose: that of portray Musk and his initiatives because the harbingers of fact.
I’m gonna go on a limb proper now and speculate that Musk isn’t any savior.
The dangers of generative AI
Despite Elon’s unsavory antics, he’s proper that generative AI poses many dangers. However, the dangers posed will not be involved with dooming civilization. That is garbage.
Perhaps the most important downside posed by generative AI within the close to future is misinformation. Bad actors can use the expertise to create convincing and tailor-made lies that might sway public opinion, manipulate elections, and incite social unrest.
For occasion, there’s AI-based software program brazenly accessible to the general public that can be utilized to create convincing fakes, corresponding to impersonating an individual’s voice and face on video. For occasion, try this 2023 Joe Rogan podcast with Apple founder Steve Jobs, who’s been dead since 2011.
And then there’s the difficulty of generative AI basically. The tempo at which AI expertise is advancing is way quicker than our means to control it.
We’ve already seen examples of generative AI getting used to create deepfakes and unfold false data. What’s to cease TruthGPT from being utilized in an identical manner?
This is an opinion article. The creator’s views don’t essentially replicate these largely held by ZME Science’s editorial employees.