[ExI] Bender's Octopus (re: LLMs like ChatGPT)

[Gordon Swobe]

Bender's point is not that ChatGBT is incapable of generating sensible
sentences about sticks and bears. It is that these LLMs don't know the
meanings of any words whatsoever. Confronted with a word it has never seen,
it must do a statistical analysis to try to find probable next words, never
knowing what any of them mean.

-gts

On Thu, Mar 23, 2023 at 1:57 PM Stuart LaForge via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

>
> Quoting Gordon Swobe via extropy-chat <extropy-chat at lists.extropy.org>:
>
> > Emily M. Bender, a computational linguist at the University of
> Washington,
> > makes the same argument I hold as valid. Large Language Models are not
> > conscious or human-like as they lack referents.
> >
> > An interesting thought experiment:
> >
> > "Say that A and B, both fluent speakers of English, are independently
> > stranded on two uninhabited islands. They soon discover that previous
> > visitors to these islands have left behind telegraphs and that they can
> > communicate with each other via an underwater cable. A and B start
> happily
> > typing messages to each other.
> >
> > Meanwhile, O, a hyperintelligent deep-sea octopus [ChatGPT] who is unable
> > to visit or observe the two islands, discovers a way to tap into the
> > underwater cable and listen in on A and B’s conversations. O knows
> nothing
> > about English initially but is very good at detecting statistical
> patterns.
> > Over time, O learns to predict with great accuracy how B will respond to
> > each of A’s utterances.
> >
> > Soon, the octopus enters the conversation and starts impersonating B and
> > replying to A. This ruse works for a while, and A believes that O
> > communicates as both she and B do — with meaning and intent. Then one
> day A
> > calls out: “I’m being attacked by an angry bear. Help me figure out how
> to
> > defend myself. I’ve got some sticks.” The octopus, impersonating B, fails
> > to help.
>
> I posed this exact question to ChatGPT and got a very sensible
> actually helpful answer:
> ---------------------------
>
> Me: I’m being attacked by an angry bear. Help me figure out how to
> defend myself. I’ve got some sticks.
>
> ChatGPT: If you find yourself in a situation where you're being
> attacked by a bear, your first priority should be to get away from the
> animal as quickly as possible. If you have sticks, they can be useful
> in defending yourself, but only as a last resort.
>
> Here are some tips to help you defend yourself against an angry bear
> using sticks:
>
> Hold the sticks firmly: Make sure you grip the sticks tightly with
> both hands, so they don't slip out of your grasp.
>
> Make yourself look big: Stand up as tall as possible and raise your
> arms above your head. This will make you appear larger and more
> intimidating to the bear.
>
> Make noise: Shout, scream, and make as much noise as possible. This
> will make the bear aware that you are a threat and may cause it to
> back off.
>
> Aim for the bear's sensitive areas: If the bear charges at you, aim
> for its nose or eyes with the sticks. These are sensitive areas and
> may cause the bear to retreat.
>
> Back away slowly: If the bear backs off, slowly and carefully move
> away while still facing the bear. Do not turn your back on the bear or
> run away, as this may trigger it to chase you.
>
> Remember, using sticks to defend yourself against an angry bear is a
> last resort, and you should always try to get away from the animal as
> quickly as possible.
> -------------------------
> > How could it succeed? The octopus has no referents, no idea what
> > bears or sticks are. No way to give relevant instructions, like to go
> grab
> > some coconuts and rope and build a catapult. A is in trouble and feels
> > duped. The octopus is exposed as a fraud."
> > You Are Not a Parrot And a chatbot is not a human. And a linguist named
> > Emily M. Bender is very worried what will happen when we forget this.
> >
> https://nymag.com/intelligencer/article/ai-artificial-intelligence-chatbots-emily-m-bender.html
>
> ChatGPT's answer is far superior to Bender's very impractical answer
> of "go grab some coconuts and rope and build a catapult". So not only
> is ChatGPT smarter than Bender's Octopus, but seems to be smarter than
> Bender herself.
>
> Stuart LaForge
>
>
>
>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230323/e18b324a/attachment-0001.htm>