29
Leak confirms OpenAI's ChatGPT will integrate MCP (www.bleepingcomputer.com)
submitted 2 days ago* (last edited 2 days ago) by Quexotic@beehaw.org to c/technology@beehaw.org
9 comments fedilink hide all child comments

An interesting development, but I doubt it'll be a good thing, especially at first. This looks like the kind of thing that will be an entirely new threat vector and a huge liability, even when used in the most secure way possible, but especially when used in a haphazard way that we'll certainly see from some of the early adoptors.

Just because you can do a thing, does not mean that you should.

I almost feel like this should have an NSFW tag because this will almost certainly not be safe for work.

Edit: looks like the article preview is failing to load... I'll try to fix it. ... Nope. Couldn't fix.

top 9 comments
sorted by: hot top controversial new old
[-] jarfil@beehaw.org 9 points 2 days ago* (last edited 2 days ago)

The connectors are still optional.

Haphazard code is not a new thing. Some statistics claim that almost 50% of "vibe coded" websites have security flaws. It's not much different from the old "12345" password, or the "qwerty" one (not naming names, but have known people using it on government infrastructure), or the "who'd want to hack us?" attitude.

MCP is the right step forward, nothing wrong with it on itself.

People disregarding basic security practices... will suffer, as always... and I don't really see anything wrong with that either. Too bad for those forced to rely on them, but that's a legislative and regulatory issue, vote accordingly.

I would still be extremely hesitant of enabling any MCP connector on non-local model instances. People need to push harder for local and on-prem AI, it's the only sane way forward.

[-] Dark_Arc@social.packetloss.gg 5 points 2 days ago* (last edited 2 days ago)

I really think we just need to move on from this AI craze.

We don't have a general intelligence. We may never have a general intelligence.

Keep using AI for what it's good for: statistics based decision making. Stop trying to use AI for designing solutions; it's not built for that because that requires reasoning which is something AI cannot do no matter how much snake oil society has been sold.

You want to use it for generating a picture, a poem, or a song ... fine, it's at least good at that because it doesn't have to solve anything using facts, making stuff up IS the goal.

[-] jarfil@beehaw.org 5 points 2 days ago* (last edited 2 days ago)

"AI" has been a buzzword basically forever, it's a moving target of "simulates some human behavior". Every time it does that, we call it an "algorithm" and move the goalpost for "true AI".

I don't know if we'll ever get AGI, or even want to, or be able to tell if we get a post-AGI. Right now, "AI" stands for something between LLMs, and Agents with an LLM core. Agents benefit from MCP, so that's good for ~~AI~~ Agents.

We can offload some basic reasoning tasks to an LLM Agent, MCP connectors allow them to interact with other services, even other agents. A lot of knowledge is locked in the deep web, and in corporate knowledge bases. The way to access those safely, will be through agents deciding which knowledge to reveal. MCP is aiming to become the new web protocol for "AI"s, no less no more.

Some careless people will get burned, the rest will be fine.

[-] Dark_Arc@social.packetloss.gg 8 points 2 days ago

We can offload some basic reasoning tasks to an LLM Agent

No, you can't. It cannot reason. It's just been fed so much existing text that it appears like it can in some cases. That's an extremely dangerous foundation on which to build anything.

[-] jarfil@beehaw.org 3 points 1 day ago* (last edited 1 day ago)

I doubt it's been fed text about "bergro", "parava", and "rortx", this looks like basic reasoning to me:

For the sake of completeness, this is qwen3:1.7b running on ollama on a smartphone. Its reasoning is more convoluted (and slow), yet the conclusion is the same:

If all bergro are rortx, and all parava are rortx, are all rortx parava?

think

Okay, let's see. The user is asking if all rortx are parava, given that all bergro are rortx and all parava are rortx. Hmm, so first, I need to parse the statements correctly.

The first part says "If all bergro are rortx, and all parava are rortx, are all rortx parava?" So, the premises are that all bergro are rortx, and all parava are rortx. The conclusion is that all rortx are parava.

Wait, but the question is a bit confusing. Let me break it down. The premises are:

  1. All bergro are rortx.
  2. All parava are rortx.

Conclusion: All rortx are parava.

So, the conclusion is that every rortx is a parava. But the premises only say that some things (bergro and parava) are rortx. But there's no information about the relationship between bergro and parava. Are bergro and parava different sets? Or is there an overlap?

If bergro and parava are distinct groups, then the premises only state that bergro are rortx and parava are rortx. But that doesn't necessarily mean that rortx are parava. Because rortx could be a subset of bergro or a separate set. For example, if bergro is a subset of rortx, and parava is another subset of rortx, then rortx would have two subsets, but the conclusion would be that all rortx are parava, which would require that rortx is entirely parava. But unless there's a statement that rortx is a subset of parava, that conclusion wouldn't hold.

But wait, the premises don't say that bergro and parava are related. They just state that all bergro are rortx and all parava are rortx. So, if bergro and parava are separate, then rortx could have elements from both. So, the conclusion that all rortx are parava would require that there's no bergro in rortx. But the premises don't say that. So, unless there's an implicit assumption that bergro and parava are disjoint, which isn't stated here, the conclusion can't be drawn.

Alternatively, if bergro and parava are the same set, but that's not stated. The problem is that the premises don't establish any relationship between bergro and parava. So, the conclusion that all rortx are parava can't be logically derived from the given premises. Therefore, the answer would be no, not all rortx are parava, unless there's additional information linking bergro and parava.

The premises state that all bergro are rortx and all parava are rortx. However, there is no information linking bergro and parava. If bergro and parava are distinct groups, rortx could consist of elements from both, making it impossible to conclude that all rortx are parava. Without additional constraints, the conclusion cannot be logically derived.

Answer: No, not all rortx are parava. The premises do not establish a relationship between bergro and parava, so rortx could include elements from both groups.

[-] Dark_Arc@social.packetloss.gg 2 points 1 day ago

Yeah, it looks like basic reasoning but it isn't. These things are based on pattern recognition. "Assume all x are y, all z are y, are all z x?" is a known formulation ... I've seen it a fair number of times in my life.

Recent development has added this whole "make it prompt itself about the question" phase to try and make things more accurate ... but that also only works sometimes.

AI in LLM form is just a sick joke. It's like watching a magic trick where half of people expect the magician to ACTUALLY levitate next year because ... "they're almost there!!"

Maybe I'll be proven wrong, but I don't see it...

[-] Quexotic@beehaw.org 2 points 2 days ago

You're not wrong, but I don't think you're 100% correct either. The human mind is able to synthesize reason by using a neural network to make connections and develop a profoundly complex statistical model using neurons. LLMs do the same thing, essentially, and they do it poorly in comparison. They don't have the natural optimizations we have, so they kinda suck at it now, but to dismiss the capabilities they currently have entirely is probably a mistake.

I'm not an apologist, to be clear. There is a ton of ethical and moral baggage tied up with the way they were made and how they're used and it needs addressed, andI think that we're only a few clever optimizations away from a threat.

[-] Dark_Arc@social.packetloss.gg 1 points 1 day ago* (last edited 1 day ago)

I don't buy the "it's a neural network" argument. We don't really understand consciousness or thinking ... and consciousness is possibly a requirement for actual thinking.

Frankly, I don't think thinking in humans is based anywhere near statical probabilities.

You can of course apply statistics and observe patterns and mimic them, but coorilation is not causation (and generally speaking, society is far too willing to accept coorilation).

Maybe everything reduces to "neural networks" in the same way LLM AI models them ... but that seems like an exceptionally bold claim for humanity to make.

[-] Quexotic@beehaw.org 1 points 21 hours ago

It makes sense that you don't buy it. LLMs are built on simplified renditions of neural structure. They're totally rudimentary.

this post was submitted on 16 May 2025
29 points (100.0% liked)

Technology

38654 readers
34 users here now

A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.

Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.

Subcommunities on Beehaw:

This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.

founded 3 years ago
MODERATORS
alyaza@beehaw.org
TheRtRevKaiser@beehaw.org
gyrfalcon@beehaw.org
rs5th@beehaw.org
coldredlight@beehaw.org
SemioticStandard@beehaw.org
TheRtRevKaiser@kbin.social
remington@beehaw.org