For librarians, educators, teachers.

State of AI right now

Technical Progress

What’s Real

• Large language models (LLMs) like GPT-4, Claude 3.5, and Gemini continue to improve in performance, speed, and usability.
• AI is now reliably useful in areas like:
  • Text summarization, content generation
  • Writing code
  • Language translation
  • Some scientific modeling (e.g., protein folding)

But:

• Breakthroughs are slowing compared to the 2022–2023 period.
• Most advances are engineering scale-ups, not fundamentally new ideas.
• Real challenges like long-term memory remain unsolved

Business & Industry

What’s Real

• AI is being used in real workflows:
  • Automating customer service
  • Speeding up legal and financial document processing
  • Assisting in education, marketing, and coding
• Enterprise adoption is high: Most large companies are integrating AI in some form.

But:

• A lot of AI adoption is shallow or experimental—especially in smaller companies.
• Many AI tools don’t justify their cost.
• Hallucinations (false responses) are still common, and human oversight is still required.