Apple defaults to OpenAI as the default search engine.
Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
11–20 of 28 posts
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#12Some random predictions: - OpenAI outages will cause more companies to explore self-hostable models - US public education buys into AI tutoring or other services. A teacher will say "just ask your AI..." to their student. - Meta's Metaverse becomes interesting instead of cringe - People still remain confused on whether to call it X or Twitter
Your first point is interesting. I’ve been meaning to explore some of these options lately (llama.cpp). I’m a bit too smooth brained to build a good business offering around that.
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#13Quantized Inertia is proven when the Barry-1 satellite[1] successfully shifts its orbit using a purely electric thruster (i.e. no propellant)
The laws of physics get re-written a bit, and Mike McCulloch starts looking like he'll get a Nobel prize out of it.
[1] https://db.satnogs.org/satellite/QDDY-8878-5291-1819-3935#ma...
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#14A weak form of AGI that while not skynet does a passible job at most tasks.
That seems less likely than rapidly self improving AI, or AI that just doesn't work too good. Why do you think AI will be able to do most tasks, but not improve itself? Unless you have a different understanding of "most tasks" than I do.
Our current method for making AI is largely a dataset creation problem.
An AI can’t just make meaningful data out of nowhere. It needs to have seen that data before.
As it stands, it can’t improve itself since it’s ability is restricted to what data it has seen.
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#15Quantized Inertia is proven when the Barry-1 satellite[1] successfully shifts its orbit using a purely electric thruster (i.e. no propellant) The laws of physics get re-written a bit, and Mike McCulloch starts looking like he'll get a Nobel prize out of it. [1] https://db.satnogs.org/satellite/QDDY-8878-5291-1819-3935#ma...
This seems really cool. Am I correct in reading that this explains away the need for (some? all?) dark matter?
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#16openai will find a reasonable model path between establishing and accumulating its own competitive barriers and building a win-win and mutually beneficial ecosystem with more entrepreneurs and companies.
Rather than easily breaking through the projects mounted on it,
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#17A weak form of AGI that while not skynet does a passible job at most tasks.
That seems less likely than rapidly self improving AI, or AI that just doesn't work too good. Why do you think AI will be able to do most tasks, but not improve itself? Unless you have a different understanding of "most tasks" than I do.
I do think it’ll improve itself. Just not singularity like fast.
A bit like a child learning doesn’t make it an Einstein overnigh
Eventually it’ll pick up pace but not in 2024
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#18- almost all AI startups based on OpenAI will fail (they have no defensibility)
- economic activity of LLM-powered self-governing agents seen in the wild
- copyright reforms lobbied by OpenAI and similar companies
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#19Convincing AI with a human voice and evil script is used for scamming, at-scale.
Conversational AI robocalls.
AI finds its way into children's pets--a next-level tamagotchi.
A company releases AI-powered talking pets for the elderly/those with alzheimers. It seems both terribly kind and cruel.
Some terribly formed AI regulation passes.
AI can create whole movies from scripts, and scripts from suggestions.
LLM-based AI's get better at chess.
AI assistant's start to get decent.
Re: Ask HN: Tech Forecasts for 2024 – Any Surprises on the Horizon?
#20Rewind.ai or Adept.ai will have monopoly over the AI application-layer market