Putting AI back in the bag

I wrote a very “stream-of-conciousness” thread on AI and how we probably can’t put it back in pandoras box on Mastodon. You can read it as a primer below:

I wanted to elaborate a bit in a longer format where I didn’t have to worry about splitting up my posts.

Generative AI as a power for good

I don’t believe that the people working on generative AI aren’t doing it out of malice. It’s very easy to see there are huge possibilities for good to come from it.

Here’s a few things I can think of off the top of my head:

• Automate away and assist in generating written communications

• Making otherwise inaccessible services accessible to this with different abilities (think of grocery shopping using something like Chat-GPT plugins).

• Solid “chat” style interfaces for customer support, brainstorming, etc.

• Generating a rough creative output with nothing but words as a starting point.

• A completely autonomous “assistant” that can work through a todo list of digital tasks for you.

• Writing, maintaining, and fixing code with a loose guide (GitHub Copilot X)

• Providing a plain English “API” to more complicated systems (Microsoft 365 Copilot)

Generative AI as a power for evil

While these are great, when you look at the list of unarguably terrible things AI can do without proper guard rails, they lose their sheen.

Here’s a few of the things I’ve been thinking about recently:

• Anyone can now make an automated, self-replying, convincing, phishing scam that responds by itself and doesn’t “believe” it’s phishing. This already exists

• Propagandists can realistically make the “dead-internet” a reality and fill faux social media sites with propaganda that will defend it’s position automatically like a real person.

• Anyone can generate completely fake photos and videos of whoever (politicians, celebrities, a family member with lots of photos online) saying and doing whatever they want. This already exists and isn’t hard to find; I’m not linking to it for hopefully obvious reasons.

• Objective truth on the digital internet will effectively be dead once these generative AI tools get sufficiently indistinguishable from reality

All very fun and very fresh ✌️

This really feels like a digital equivalent to CFCs burning a hole in our o-zone layer; luckily we were able to regulate their use and prevent the earth from crisping itself.

Hopefully we learn to get our collective minds out of the clouds and start regulating this, before we lose the internet as we know it today.

Responses

This post was liked by 3 people

• 
  leadb in English

  @mykalmachon

  I’m 58yo and was thinking the same some 30 years ago about digital (computers, Internet, web services… and ai).

  As you sort of mentioned in another post, the toxic uses of each tech seam to become the norm while uses respecting users and their living ecosystem are just shown to entertain quiproquos.

  i.e. even when they are proliferating, they remain a very small part compared to the toxic norms.

  ex of quiproquo? « For good » without saying for whom.

  🤓

  received on Sun Mar 26 2023

Responses on this site are powered by Webmentions!