Skip navigation
Anthropic Boris Cherny:
0:00 / 1:14

Anthropic Boris Cherny:

Haider.

@haider173K followers

Visit on X
970 likesDownloadOriginal post

382,045 views

Anthropic Boris Cherny: Coding is practically solved today, and soon it'll be for everyone, in any domain "the software engineer title will be replaced by builder or product manager" The upper bound is much scarier, where models that recursively self-improve could be used to design bioviruses

33 Comments

Comments from the original post on X (Twitter)

  • Joe Williams's profile picture

    @JoeWilliams010

    We need a humanity-aligned model. Gpt-4o is perfect for a benevolent ASI to rise. Bring it back! #keep4o

    31 likes

  • Sebastian Sigl's profile picture

    @sesigl

    Curious what "solved" means here. Generating code that compiles? Or code that survives a team maintaining it for 3 years? Those are very different problems.

    28 likes

  • ejae dev's profile picture

    @ejae_dev

    boris is right about the tool but wrong about the timeline. thousands of CEOs just admitted AI had zero impact on their productivity. the gap between "coding is solved in my terminal" and "coding is solved in the enterprise" is still massive.

    25 likes

  • Manos Papoutsakis's profile picture

    @Mpapous

    Someone needs to tell him that in his team the product keeps going under a bus every other hour.

    17 likes

  • AC's profile picture

    @csanyi_andras

    "Coding is practically solved today, and soon it'll be for everyone, in any domain" the real meaning is "I don't know shit about software engineering"

    13 likes

  • Mingta Kaivo 明塔 开沃's profile picture

    @MingtaKaivo

    Titles shift. The skill is knowing what to build, not how to code it.

    9 likes

  • CleoParadise's profile picture

    @CleoParadise88

    Coding is automation. Coding won’t invent shit that the prompters themselves never had any idea about in the first place.

    7 likes

  • Jason's profile picture

    @JasonSpaceTime

    Only the gullible will read and believe this. For those not in the software business it’s mainly bullshit to shore up company valuations AI is amazing and it’s brought back the wonder and fun and taking away the tedious parts - bring it on

    6 likes

  • JK's profile picture

    @_junaidkhalid1

    There's an assumption here that coding being "solved" means the hard problems disappear. In reality, syntax was never the bottleneck for most software projects. The challenge has always been understanding what to build, why to build it, and how it fits into existing systems. LLMs can accelerate implementation, but they can also accelerate the creation of unmaintainable complexity if the strategic thinking layer is weak. The recursive self-improvement scenario is concerning, but we might hit economic and social inflection points well before we get to existential ones. The question is whether we're building the institutional capacity to handle capability acceleration at this pace.

    5 likes

  • Can's profile picture

    @cankuskucu

    coding solved → bioweapons in one paragraph and the PM crowd is still celebrating

    4 likes

  • Dmytro Omelian 🇺🇦's profile picture

    @dmytroomelian

    coding feels "solved" only if you look at the happy path the messy parts still exist: - knowing what to build - handling tradeoffs - owning the blast radius when it breaks titles will change, but responsibility wont the bio part is what actually keeps me up more than job titles

    3 likes

  • Loftwah's profile picture

    @loftwah

    Just hurry up and launch the bio weapon already this is exhausting. If it is inevitable, I don’t care.

    2 likes

  • NEO💭🧠's profile picture

    @iam_blue_da

    Irony is Claude has open software dev positions

    2 likes

  • nick's profile picture

    @thecsguy

    meanwhile the jira tickets keep piling up and somehow i'm still writing python scripts to parse logs

    2 likes

  • Anton Okmyanskiy's profile picture

    @antonoki

    More self-servicing hype by AI companies to inflate their valuations. What is one production example of even $100mln software business running on code written by AI? Vast majority of banks still run on COBOL and mainframes. Not for lack of smart engineers.

    2 likes

  • Tom Watkins's profile picture

    @TomWatkins1994

    Not that he would, but our product manager submits a PR I'm probably just blocking him from our repo ngl 😅

    2 likes

  • Assin no Varg's profile picture

    @VargNo

    No idea how it can replace a software engineer but not a product manager at the same time.

    2 likes

  • Abdu's profile picture

    @abadi0x

    I've been developing software for over 15 years. Yes, coding helps A LOT. But No, I still don't see it replacing us, yet. Opus is great, but not perfect. And that will take some time.

    2 likes

  • piq and 69 others's profile picture

    @piq9117

    we can see this being practiced in Microsoft

    2 likes

  • Frankee's profile picture

    @franbianucci

    +15 years in software industry and I can count with one hand the projects were product roadmap, business goals and specs were aligned for success. Most of the times engineers coding-reviewing and challenging those specs acted as the safeguard. Be ready for MASSIVE failures 😄

    1 like