WHY CAN YOU LOSE BY INVESTING IN AI? ๐ค๐ธ
#ChatGPT is a before and after in the economy. Many sectors are now disrupted.
This should greatly increase investment in AI versus the old-school version of doing things.
But not all #AI #investment will succeed.
๐งต
#ChatGPT is a before and after in the economy. Many sectors are now disrupted.
This should greatly increase investment in AI versus the old-school version of doing things.
But not all #AI #investment will succeed.
๐งต
Here's my take on the two divisions of AI (feedback welcome):
THE SATURN'S SON:
AIs based on PUBLIC models or data.
Startups that use models like #GPT or others based on image, text... trained with publicly available dataset/models, maybe adding their own "layers/fine tuning".
THE SATURN'S SON:
AIs based on PUBLIC models or data.
Startups that use models like #GPT or others based on image, text... trained with publicly available dataset/models, maybe adding their own "layers/fine tuning".
The part they are building, in many cases, may be overperformed for the biggest models, their "parents".
Like Saturn and his Son.
Like Saturn and his Son.
Examples: content creators, legal texts, customer supports, image based generative AI... Many of them are going to be surpassed by their "father" #ChatGPT. ๐
Similar situation for generic image generators. ๐ผ
This is also a case for #blueocean Ref: en.wikipedia.org/wiki/Blue_Oceaโฆ
Similar situation for generic image generators. ๐ผ
This is also a case for #blueocean Ref: en.wikipedia.org/wiki/Blue_Oceaโฆ
THE TRACEUR
AI based on SPECIFIC models and SPECIFIC data.
This models are designed for specific domains and use data that is not available for the public or big players and understand their domains and nature better than generic AIs.
AI based on SPECIFIC models and SPECIFIC data.
This models are designed for specific domains and use data that is not available for the public or big players and understand their domains and nature better than generic AIs.
Instead of training to be faster than Usain Bolt ๐โโ๏ธ on a track, you want to be a traceur who can navigate a different path using specific techniques and feedback.
If you prefer, you want to be a 4x4 ๐ that transport your passengers over steep paths where a bus ๐ cannot go.
If you prefer, you want to be a 4x4 ๐ that transport your passengers over steep paths where a bus ๐ cannot go.
HOW TO BE A TRACEUR:
You must be proficient solving problems in a domain large enough but not yet covered to get the leadership. ๐ฅ
Instead of using #AI models like a hammer for every task, you should decode the problem and design a solution that fits the nature of the field.
You must be proficient solving problems in a domain large enough but not yet covered to get the leadership. ๐ฅ
Instead of using #AI models like a hammer for every task, you should decode the problem and design a solution that fits the nature of the field.
Be ready to gather the most purest data from that field under your structure.
If possible, supplement your #AI with "generic vitamins" from generic AIs, but as an addon.
Remember to protect your core, your #competitiveadvantage, your secret sauce, your #moat.
If possible, supplement your #AI with "generic vitamins" from generic AIs, but as an addon.
Remember to protect your core, your #competitiveadvantage, your secret sauce, your #moat.
And if you are an #investor #angel #vc and want to know more "traceur AI #startups" or are looking in any specific vertical, feel free to DM us to see if we can send you some.
Also tell you that we are now #fundarising, in case you want to talk to see if there is a good fit.
Also tell you that we are now #fundarising, in case you want to talk to see if there is a good fit.
@threadreaderapp unroll
โข โข โข
Missing some Tweet in this thread? You can try to
force a refresh
ใ
